From 6ad5c0bac0841ff649c64446472be11b86be641a Mon Sep 17 00:00:00 2001
From: wlandau <will.landau.oss@gmail.com>
Date: Wed, 29 May 2024 16:27:18 -0400
Subject: [PATCH] brm_recenter_nuisance()

---
 NAMESPACE                                     |   1 +
 R/brm_recenter_nuisance.R                     |  80 +++++++++++++++++
 R/utils_archetype.R                           |  19 ++--
 man/brm_prior_archetype.Rd                    |  12 +--
 man/brm_prior_label.Rd                        |   7 +-
 man/brm_recenter_nuisance.Rd                  |  74 ++++++++++++++++
 tests/testthat/test-brm_recenter_nuisance.R   |  37 ++++++++
 vignettes/archetypes.Rmd                      |  82 +++++++++---------
 vignettes/archetypes.Rmd.upstream             |   2 +-
 .../archetype_compare_data-1.png              | Bin 16253 -> 16304 bytes
 10 files changed, 251 insertions(+), 63 deletions(-)
 create mode 100644 R/brm_recenter_nuisance.R
 create mode 100644 man/brm_recenter_nuisance.Rd
 create mode 100644 tests/testthat/test-brm_recenter_nuisance.R

diff --git a/NAMESPACE b/NAMESPACE
index 0fbe7ffb..956d1a9e 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -37,6 +37,7 @@ export(brm_plot_draws)
 export(brm_prior_archetype)
 export(brm_prior_label)
 export(brm_prior_simple)
+export(brm_recenter_nuisance)
 export(brm_simulate)
 export(brm_simulate_categorical)
 export(brm_simulate_continuous)
diff --git a/R/brm_recenter_nuisance.R b/R/brm_recenter_nuisance.R
new file mode 100644
index 00000000..f223687f
--- /dev/null
+++ b/R/brm_recenter_nuisance.R
@@ -0,0 +1,80 @@
+#' @title Recenter nuisance variables
+#' @export
+#' @family archetype utilities
+#' @description Change the center of a nuisance variable of an
+#'   informative prior archetype.
+#' @details By "centering vector y at scalar x", we mean taking
+#'   the difference `z = y - x`. If `x` is the mean, then `mean(z)` is
+#'   0. Informative prior archetypes center nuisance variables
+#'   at their means so the parameters can be interpreted correctly
+#'   for setting informative priors. This is appropriate most of the time,
+#'   but sometimes it is better to center a column at a pre-specified
+#'   scientifically meaningful fixed number. If you want a nuisance column
+#'   to be centered at a fixed value other than its mean,
+#'   use [brm_recenter_nuisance()] to shift the center. This function
+#'   can handle any nuisance variable
+#' @return An informative prior archetype data frame with one of the
+#'   variables re-centered.
+#' @param data An informative prior archetype data frame output from
+#'   [brm_archetype_cells()] or similar.
+#' @param nuisance Character of length 1, name of the nuisance column
+#'   in the data to shift the center.
+#' @param center Numeric of length 1, value of the center to shift
+#'   the column in `nuisance`. The affected column in the returned
+#'   archetype data frame will look as if it were centered by this
+#'   value to begin with.
+#' @examples
+#' set.seed(0L)
+#' data <- brm_simulate_outline(
+#'   n_group = 2,
+#'   n_patient = 100,
+#'   n_time = 4,
+#'   rate_dropout = 0,
+#'   rate_lapse = 0
+#' ) |>
+#'   dplyr::mutate(response = rnorm(n = dplyr::n())) |>
+#'   brm_data_change() |>
+#'   brm_simulate_continuous(names = c("biomarker1", "biomarker2")) |>
+#'   brm_simulate_categorical(
+#'     names = c("status1", "status2"),
+#'     levels = c("present", "absent")
+#'   )
+#' archetype <- brm_archetype_cells(data)
+#' mean(archetype$nuisance_biomarker1) # after original centering
+#' center <- mean(data$biomarker1)
+#' center # original center, before the centering from brm_archetype_cells()
+#' attr(archetype$nuisance_biomarker1, "brm_center") # original center
+#' max(abs((data$biomarker1 - center) - archetype$nuisance_biomarker1))
+#' # Re-center nuisance_biomarker1 at 0.75.
+#' archetype <- brm_recenter_nuisance(
+#'   data = archetype,
+#'   nuisance = "nuisance_biomarker1",
+#'   center = 0.75
+#' )
+#' attr(archetype$nuisance_biomarker1, "brm_center") # new center
+#' mean(archetype$nuisance_biomarker1) # no longer equal to the center
+#' # nuisance_biomarker1 is now as though we centered it at 0.75.
+#' max(abs((data$biomarker1 - 0.75) - archetype$nuisance_biomarker1))
+brm_recenter_nuisance <- function(data, nuisance, center) {
+  brm_data_validate(data = data)
+  assert(
+    inherits(data, "brms_mmrm_archetype"),
+    message = "data must be a brms.mmrm informative prior archetype"
+  )
+  assert_chr(nuisance, "nuisance must be a character string")
+  assert(
+    nuisance %in% attr(data, "brm_archetype_nuisance"),
+    message = c(
+      "nuisance must be the name of a nuisance column in the archetype data"
+    )
+  )
+  assert_num(center, "center must be a valid non-missing numeric scalar")
+  original <- attr(data[[nuisance]], "brm_center")
+  assert_num(
+    original,
+    paste("original center of", nuisance, "could not be found")
+  )
+  data[[nuisance]] <- data[[nuisance]] + original - center
+  attr(data[[nuisance]], "brm_center") <- center
+  data
+}
diff --git a/R/utils_archetype.R b/R/utils_archetype.R
index 775b5b12..4809e3cd 100644
--- a/R/utils_archetype.R
+++ b/R/utils_archetype.R
@@ -49,7 +49,7 @@ archetype_nuisance <- function(
       nuisance = out
     )
   }
-  out
+  nuisance_center(out)
 }
 
 nuisance_covariates <- function(data) {
@@ -62,16 +62,11 @@ nuisance_covariates <- function(data) {
     colnames(categorical) <- paste0(colnames(categorical), "_")
     out <- dplyr::bind_cols(out, model.matrix(~ 0 + ., categorical))
   }
-  for (name in colnames(out)) {
-    out[[name]] <- out[[name]] - mean(out[[name]])
-  }
   out
 }
 
 nuisance_baseline <- function(data) {
-  baseline <- attr(data, "brm_baseline")
-  data[[baseline]] <- data[[baseline]] - mean(data[[baseline]])
-  data[, baseline, drop = FALSE]
+  data[, attr(data, "brm_baseline"), drop = FALSE]
 }
 
 nuisance_baseline_subgroup <- function(data) {
@@ -98,7 +93,6 @@ nuisance_baseline_time <- function(data) {
 
 nuisance_baseline_terms <- function(data, baseline, formula) {
   matrix <- model.matrix(object = formula, data = data)
-  matrix <- sweep(matrix, MARGIN = 2L, STATS = colMeans(matrix), FUN = "-")
   tibble::as_tibble(as.data.frame(matrix))
 }
 
@@ -136,3 +130,12 @@ drop_zero_columns <- function(x) {
   sums <- colSums(abs(x))
   x[, sums > .Machine$double.eps, drop = FALSE]
 }
+
+nuisance_center <- function(data) {
+  for (name in colnames(data)) {
+    mean <- mean(data[[name]])
+    attr(data[[name]], "brm_center") <- mean
+    data[[name]] <- data[[name]] - mean
+  }
+  data
+}
diff --git a/man/brm_prior_archetype.Rd b/man/brm_prior_archetype.Rd
index 340f217c..5df36248 100644
--- a/man/brm_prior_archetype.Rd
+++ b/man/brm_prior_archetype.Rd
@@ -61,18 +61,14 @@ data <- brm_simulate_outline(
   )
 archetype <- brm_archetype_successive_cells(data)
 dplyr::distinct(data, group, time)
-label <- brm_prior_label(
-  code = "normal(1, 1)",
-  group = "group_1",
-  time = "time_1"
-) |>
+prior <- NULL |>
+  brm_prior_label("normal(1, 1)", group = "group_1", time = "time_1") |>
   brm_prior_label("normal(1, 2)", group = "group_1", time = "time_2") |>
   brm_prior_label("normal(1, 3)", group = "group_1", time = "time_3") |>
   brm_prior_label("normal(2, 1)", group = "group_2", time = "time_1") |>
   brm_prior_label("normal(2, 2)", group = "group_2", time = "time_2") |>
-  brm_prior_label("normal(2, 3)", group = "group_2", time = "time_3")
-label
-prior <- brm_prior_archetype(label = label, archetype = archetype)
+  brm_prior_label("normal(2, 3)", group = "group_2", time = "time_3") |>
+  brm_prior_archetype(archetype = archetype)
 prior
 class(prior)
 }
diff --git a/man/brm_prior_label.Rd b/man/brm_prior_label.Rd
index ddde3552..7f711970 100644
--- a/man/brm_prior_label.Rd
+++ b/man/brm_prior_label.Rd
@@ -80,11 +80,8 @@ data <- brm_simulate_outline(
   )
 archetype <- brm_archetype_successive_cells(data)
 dplyr::distinct(data, group, time)
-label <- brm_prior_label(
-  code = "normal(1, 1)",
-  group = "group_1",
-  time = "time_1"
-) |>
+label <- NULL |>
+  brm_prior_label("normal(1, 1)", group = "group_1", time = "time_1") |>
   brm_prior_label("normal(1, 2)", group = "group_1", time = "time_2") |>
   brm_prior_label("normal(1, 3)", group = "group_1", time = "time_3") |>
   brm_prior_label("normal(2, 1)", group = "group_2", time = "time_1") |>
diff --git a/man/brm_recenter_nuisance.Rd b/man/brm_recenter_nuisance.Rd
new file mode 100644
index 00000000..c7f87429
--- /dev/null
+++ b/man/brm_recenter_nuisance.Rd
@@ -0,0 +1,74 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/brm_recenter_nuisance.R
+\name{brm_recenter_nuisance}
+\alias{brm_recenter_nuisance}
+\title{Recenter nuisance variables}
+\usage{
+brm_recenter_nuisance(data, nuisance, center)
+}
+\arguments{
+\item{data}{An informative prior archetype data frame output from
+\code{\link[=brm_archetype_cells]{brm_archetype_cells()}} or similar.}
+
+\item{nuisance}{Character of length 1, name of the nuisance column
+in the data to shift the center.}
+
+\item{center}{Numeric of length 1, value of the center to shift
+the column in \code{nuisance}. The affected column in the returned
+archetype data frame will look as if it were centered by this
+value to begin with.}
+}
+\value{
+An informative prior archetype data frame with one of the
+variables re-centered.
+}
+\description{
+Change the center of a nuisance variable of an
+informative prior archetype.
+}
+\details{
+By "centering vector y at scalar x", we mean taking
+the difference \code{z = y - x}. If \code{x} is the mean, then \code{mean(z)} is
+0. Informative prior archetypes center nuisance variables
+at their means so the parameters can be interpreted correctly
+for setting informative priors. This is appropriate most of the time,
+but sometimes it is better to center a column at a pre-specified
+scientifically meaningful fixed number. If you want a nuisance column
+to be centered at a fixed value other than its mean,
+use \code{\link[=brm_recenter_nuisance]{brm_recenter_nuisance()}} to shift the center. This function
+can handle any nuisance variable
+}
+\examples{
+set.seed(0L)
+data <- brm_simulate_outline(
+  n_group = 2,
+  n_patient = 100,
+  n_time = 4,
+  rate_dropout = 0,
+  rate_lapse = 0
+) |>
+  dplyr::mutate(response = rnorm(n = dplyr::n())) |>
+  brm_data_change() |>
+  brm_simulate_continuous(names = c("biomarker1", "biomarker2")) |>
+  brm_simulate_categorical(
+    names = c("status1", "status2"),
+    levels = c("present", "absent")
+  )
+archetype <- brm_archetype_cells(data)
+mean(archetype$nuisance_biomarker1) # after original centering
+center <- mean(data$biomarker1)
+center # original center, before the centering from brm_archetype_cells()
+attr(archetype$nuisance_biomarker1, "brm_center") # original center
+max(abs((data$biomarker1 - center) - archetype$nuisance_biomarker1))
+# Re-center nuisance_biomarker1 at 0.75.
+archetype <- brm_recenter_nuisance(
+  data = archetype,
+  nuisance = "nuisance_biomarker1",
+  center = 0.75
+)
+attr(archetype$nuisance_biomarker1, "brm_center") # new center
+mean(archetype$nuisance_biomarker1) # no longer equal to the center
+# nuisance_biomarker1 is now as though we centered it at 0.75.
+max(abs((data$biomarker1 - 0.75) - archetype$nuisance_biomarker1))
+}
+\concept{archetype utilities}
diff --git a/tests/testthat/test-brm_recenter_nuisance.R b/tests/testthat/test-brm_recenter_nuisance.R
new file mode 100644
index 00000000..21a8094b
--- /dev/null
+++ b/tests/testthat/test-brm_recenter_nuisance.R
@@ -0,0 +1,37 @@
+test_that("multiplication works", {
+  set.seed(0L)
+  data <- brm_simulate_outline(
+    n_group = 2,
+    n_patient = 100,
+    n_time = 4,
+    rate_dropout = 0,
+    rate_lapse = 0
+  ) |>
+    dplyr::mutate(response = rnorm(n = dplyr::n())) |>
+    brm_data_change() |>
+    brm_simulate_continuous(names = c("biomarker1", "biomarker2")) |>
+    brm_simulate_categorical(
+      names = c("status1", "status2"),
+      levels = c("present", "absent")
+    )
+  archetype <- brm_archetype_cells(data)
+  expect_equal(mean(archetype$nuisance_biomarker1), 0)
+  expect_equal(
+    mean(data$biomarker1),
+    attr(archetype$nuisance_biomarker1, "brm_center")
+  )
+  expect_equal(
+    max(abs((data$biomarker1 - center) - archetype$nuisance_biomarker1)),
+    0
+  )
+  archetype <- brm_recenter_nuisance(
+    data = archetype,
+    nuisance = "nuisance_biomarker1",
+    center = 0.75
+  )
+  expect_equal(attr(archetype$nuisance_biomarker1, "brm_center"), 0.75)
+  expect_equal(
+    max(abs((data$biomarker1 - 0.75) - archetype$nuisance_biomarker1)),
+    0
+  )
+})
diff --git a/vignettes/archetypes.Rmd b/vignettes/archetypes.Rmd
index 7cf0f529..b44b1afd 100644
--- a/vignettes/archetypes.Rmd
+++ b/vignettes/archetypes.Rmd
@@ -100,7 +100,7 @@ attr(archetype, "brm_archetype_interest")
 #> [5] "x_group_2_time_3" "x_group_2_time_4"
 ```
 
-and we have nuisance variables. Some nuisance variables are continuous covariates, while others are levels of one-hot-encoded concomitant factors or interactions of those concomitant factors with baseline and/or subgroup. All nuisance variables are centered at their means so the reference level of the model is at the "center" of the data and not implicitly conditional on a subset of the data. In addition, some nuisance variables are automatically dropped in order to ensure the model matrix is full-rank. This is critically important to preserve the interpretation of the columns of interest and make sure the informative priors behave as expected.
+and we have nuisance variables. Some nuisance variables are continuous covariates, while others are levels of one-hot-encoded concomitant factors or interactions of those concomitant factors with baseline and/or subgroup. All nuisance variables are centered at their means so the reference level of the model is at the "center" of the data and not implicitly conditional on a subset of the data.^[`brm_recenter_nuisance()` can retroactively recenter a nuisance column to a fixed value other than its mean.] In addition, some nuisance variables are automatically dropped in order to ensure the model matrix is full-rank. This is critically important to preserve the interpretation of the columns of interest and make sure the informative priors behave as expected.
 
 
 ```r
@@ -293,46 +293,46 @@ model <- brm_model(
 #> Compiling Stan program...
 #> Start sampling
 brms::prior_summary(model)
-#>                     prior    class                         coef group resp
-#>                    (flat)        b                                        
-#>  student_t(4, 2.17, 4.86)        b nuisance_baseline.timetime_2           
-#>  student_t(4, 3.22, 5.25)        b nuisance_baseline.timetime_3           
-#>  student_t(4, 2.53, 5.75)        b nuisance_baseline.timetime_4           
-#>                    (flat)        b          nuisance_biomarker1           
-#>                    (flat)        b          nuisance_biomarker2           
-#>                    (flat)        b      nuisance_status1_absent           
-#>                    (flat)        b     nuisance_status2_present           
-#>  student_t(4, 0.98, 2.37)        b             x_group_1_time_2           
-#>  student_t(4, 1.82, 3.32)        b             x_group_1_time_3           
-#>  student_t(4, 2.35, 4.41)        b             x_group_1_time_4           
-#>  student_t(4, 0.31, 2.22)        b             x_group_2_time_2           
-#>  student_t(4, 1.94, 2.85)        b             x_group_2_time_3           
-#>  student_t(4, 2.33, 3.41)        b             x_group_2_time_4           
-#>                    (flat)        b                                        
-#>                    (flat)        b                   timetime_2           
-#>                    (flat)        b                   timetime_3           
-#>                    (flat)        b                   timetime_4           
-#>      lkj_corr_cholesky(1) Lcortime                                        
-#>   dpar nlpar lb ub       source
-#>                         default
-#>                            user
-#>                            user
-#>                            user
-#>                    (vectorized)
-#>                    (vectorized)
-#>                    (vectorized)
-#>                    (vectorized)
-#>                            user
-#>                            user
-#>                            user
-#>                            user
-#>                            user
-#>                            user
-#>  sigma                  default
-#>  sigma             (vectorized)
-#>  sigma             (vectorized)
-#>  sigma             (vectorized)
-#>                         default
+#>                     prior    class                         coef group resp  dpar nlpar
+#>                    (flat)        b                                                    
+#>  student_t(4, 2.17, 4.86)        b nuisance_baseline.timetime_2                       
+#>  student_t(4, 3.22, 5.25)        b nuisance_baseline.timetime_3                       
+#>  student_t(4, 2.53, 5.75)        b nuisance_baseline.timetime_4                       
+#>                    (flat)        b          nuisance_biomarker1                       
+#>                    (flat)        b          nuisance_biomarker2                       
+#>                    (flat)        b      nuisance_status1_absent                       
+#>                    (flat)        b     nuisance_status2_present                       
+#>  student_t(4, 0.98, 2.37)        b             x_group_1_time_2                       
+#>  student_t(4, 1.82, 3.32)        b             x_group_1_time_3                       
+#>  student_t(4, 2.35, 4.41)        b             x_group_1_time_4                       
+#>  student_t(4, 0.31, 2.22)        b             x_group_2_time_2                       
+#>  student_t(4, 1.94, 2.85)        b             x_group_2_time_3                       
+#>  student_t(4, 2.33, 3.41)        b             x_group_2_time_4                       
+#>                    (flat)        b                                         sigma      
+#>                    (flat)        b                   timetime_2            sigma      
+#>                    (flat)        b                   timetime_3            sigma      
+#>                    (flat)        b                   timetime_4            sigma      
+#>      lkj_corr_cholesky(1) Lcortime                                                    
+#>  lb ub       source
+#>             default
+#>                user
+#>                user
+#>                user
+#>        (vectorized)
+#>        (vectorized)
+#>        (vectorized)
+#>        (vectorized)
+#>                user
+#>                user
+#>                user
+#>                user
+#>                user
+#>                user
+#>             default
+#>        (vectorized)
+#>        (vectorized)
+#>        (vectorized)
+#>             default
 ```
 
 Marginal mean estimation, post-processing, and visualization automatically understand the archetype without any user intervention.
diff --git a/vignettes/archetypes.Rmd.upstream b/vignettes/archetypes.Rmd.upstream
index 6da23f10..09f98f97 100644
--- a/vignettes/archetypes.Rmd.upstream
+++ b/vignettes/archetypes.Rmd.upstream
@@ -70,7 +70,7 @@ Those new columns constitute a custom model matrix to describe the desired param
 attr(archetype, "brm_archetype_interest")
 ```
 
-and we have nuisance variables. Some nuisance variables are continuous covariates, while others are levels of one-hot-encoded concomitant factors or interactions of those concomitant factors with baseline and/or subgroup. All nuisance variables are centered at their means so the reference level of the model is at the "center" of the data and not implicitly conditional on a subset of the data. In addition, some nuisance variables are automatically dropped in order to ensure the model matrix is full-rank. This is critically important to preserve the interpretation of the columns of interest and make sure the informative priors behave as expected.
+and we have nuisance variables. Some nuisance variables are continuous covariates, while others are levels of one-hot-encoded concomitant factors or interactions of those concomitant factors with baseline and/or subgroup. All nuisance variables are centered at their means so the reference level of the model is at the "center" of the data and not implicitly conditional on a subset of the data.^[`brm_recenter_nuisance()` can retroactively recenter a nuisance column to a fixed value other than its mean.] In addition, some nuisance variables are automatically dropped in order to ensure the model matrix is full-rank. This is critically important to preserve the interpretation of the columns of interest and make sure the informative priors behave as expected.
 
 ```{r}
 attr(archetype, "brm_archetype_nuisance")
diff --git a/vignettes/archetypes_figures/archetype_compare_data-1.png b/vignettes/archetypes_figures/archetype_compare_data-1.png
index 6f27120184ff06daeb252d54fc6d68d8ed1d78e9..c60a954b982ddccfecdb0d7256aedd79336d1d2f 100644
GIT binary patch
literal 16304
zcmch82UJw;l5L|XC{c+L1yMk9&RI|-g91&?AW=Y~1O)`iNK~?bAX##TA4viNO-=#|
zA{iuSnCkzAnLF>^w`SIxdA%06ICh^t=R5UP?b>@+u&VNXysH#f5eNjH{DXVy2n5Cf
z`Um?09ASQF#RUIcdHMk9j6mRANB>|9O!=H35H}I>_oN=VC$CL;coDgrN&N1ks)?z5
zVzVU0T>jw^hO87uoUABe>xN%OI!ngKm70u#4wVsEO(X#!8A_j|O;$<@Ll#LKMG~bT
z`RL<=8fMiZ^@;2rvC)mle*47ioiDtDUPqqdNB6fn8F@D3xr-Gg-ubH$zk;i2&~Cs%
zAYP2TO2<N^Q~A*#UN~WUA`o4hC28j7<{ud3OVT)0Z}IW*6?~|#*GqW#0-vfwC`Gv;
z-%3^(v!Q$b>ScfQJ2HJxT8|zrs$W4hTk#H1?~vDG{KM(!7U#5;l<?ZKrDWI>Y|hV4
z$F%&!v5(wu%MT<AIr|*^z-qw3cxXFbIzMiicDmKgk><I~jEvu9$0`p<k9;M0^Xc#J
zQdN`Qv#nQ}W_(W$HZ$)>hGZ~>%$}Vd=jZ2Np))A1Dld;EoOP{JNfDlh->IS=JB$6L
zrKPV2z0M97s_dsI2HRd#{%j2<vg}P3@;dm$Ab<PzZJNl1gUz{ssI9FnnIQb!g-^op
zKTIJ3QK!fIMAG<Fg5P&N_tz%L9Fx2^XWKX$4`$~gB~p`<>n;0I$Jp`8+1DL>k9ET=
zk&hn_7o7?09G}MAQQ}VO*<R@5U}5R*>hk>c&N44A&vS28$Z_^w@o1g<=IqYSv*OX?
z!-cf{jhXX>G~elZkFriPPq<8P@fr9#_3hiYDJj(a<5K>Y<)h6?wBhQWA=#-vWMu_Z
zB#S-QOnA7nHJ@PUbG%l?sGR(k%d}ldU}|SX*P&3iik*$Eu)6vHE>qlPc?jO*G5lCd
zDT%*OB~6?(bRI5u5>Mb8r+HUwn$J;gL4j?v-z9%60t@mx&k_$04{zMK5zXKYfda>r
zA3Q+WESW^kRcB=}{Q1aN@6SvZ7&iI*@dBgi{B)zNw6yX3?78%V1a|$ilLZOi)7?=m
z1CM#Os&7v4kOXGDoL9$6j8#}!S-lQ^1i^A-tE5dg`TB}k2k~3Cg^)PZZ(eTUoUF9d
z&}|Q=Oh`*RN68SJZl|696kKm3GkjDv+w$%CMgtOw9L!c?VP{{TZg{>Z<Kf|vqnwf`
z?6Pd0pdTL>M-tlhLR$(|RSOGUUd}81dfZl1_c{^L6*>|Ul6UXk1w_I17mw;r8lD;^
zow#bNIysfz#O5lxF(D`)J+!v@iGYMIU`|l_fx3Eh-$MSH;qiO*^BWJ9m5c0hE8I4w
zb2W0rx4ztJc)o8xzeP8~DL1nWPXvAE$oSC$B+-O{wsy5~3p^{$Vgv8ZHZqYEWtS7F
zsb$816L=8fX%0P!d}Ml0U*fySP+IUOsU!<J&b9{6&d#D->!ym%D9UX{i%w3wat4tG
zf4<b!GhD~YpTvx=Ow~AB!s33;K3uQcX#D*dr@?bKn&C=9ZaQw|qOShNbOQmEAROi}
zGdJhLHRXOcgt2eX;7Kj!xl^-{+TGJbE#g|xY*hW-4<}nG3D)X^y{oyI*~uL3`6r!1
z0<j&{R_2@!AB3nP-m>GC<Ee;lwtlN|F4VNKw=ZfoTKra1)@*b$K=klu%=y`V(<LeQ
zpKtEl@=w}TO;tYqA)VCYI^{B&t0L<B>#G7s1Fyx`w?{I+@5C%sj2dV-NWhvwM3v~#
zTRm1>ZK_px_;6yl#vY$m{G^3IB8*DN>Fnfie0<#hGM;r;nd|zu8kd!k<BcZ6hC0`^
z3DlHBAReVf)uir+Z|<t9s)Qq@4m0BVcl)~`AlwF(1g=SEE5744Z4bL6Zu2?dibjqy
zA4PaT6m0nlo6!;Rp&I9fM(^WFxm-ucNpN=L#$BCj=tWaj?wzQxN%P(tx1XxTqZ0fS
z+*dko3xQwZmUOU-SZjTOadFyx4o_B%1~OB&1Pcd8q`5D3p^}0ErI_cll}A{1_?3%v
zXGWnk<+7)wVm@J){+8^7!}@cwv)g}t(T+Fov2b2wUv_LZDZ%t_J`i>vGvegL&76C!
z7lco3!gDLL|0B~wFY56>9FzO4aHk5z*|iHbn`UT}%GN(((W)v2rITUO@iVt?i*`6}
zRw!SWUUjWwkl%b<NwFh5LRE%LQp~Q#@s!5yE_Kx6>2S<oYQMwsQ6iW!&>n$c3i?m!
zoNf{0UmVPlLiHsHY;-cFC5w8LPOjnG?Rz#<<Y-_)ZMa9d^Wnn>$dZFODzuzNe}s^*
zNYm<+TdAF{s9!)xhRIUyw6(QCMA8TBtd2WFFjlPPsH8zmcXoEFDbpel8N_SyJ)$&`
zo}2Rd!xWyL^_Or+zd3eL!*0&E^O%p*if2ViV<X{|_V)I%+jg}>Lql(S6j*|TgFO>a
z*DcDB#mJ5W{Yss}lH`;-vyQS)g;Pr2sd=TaR$j`$JDTv?KlG{~%bPB<j2v82R8YWK
zJA@6FpCCpgySX}Erq)}<EEu8i;K9d&g0SX#-q_gKpPTNpD|-H<p|8x5ksU-y<GExi
z1SCl+#B{9|Wu2u|hB+sE<hcAvg?V`oH>49gmY!*9X?5K36N-TF4QUSBxo(FdBPI3u
z-Hg=|h{tmG?(%d)-9$B%gO(tIri0l~MwL`Z(R&vOL?yAgl6sa&4WW>-s$8Wwwz(~*
zq82O2R+sWsxN(|#Uci%Vz?O?egvK!0BA9<s#AP`)E^fD#NQL3f9r4qh!GI{ox%QqE
z5x3UY*8-x>_s`GL61NbDbTVpod;=u%Mj1{d0|UdOM~_;K3@Dt0Q|zlm8UyVMG+DY!
zjgLnwQ;%AD=I7^2-c?N5CzuO0vg?xO$OyobVNYPOkP+ZI7Lz0sG`b2EUs&&9l;_Dh
z{b0LGxT~qzJEqdlubM7h?|5f6PQH~EpFG?tU>i^=dqMKJf5d;fe9xw+zF^bGPf|O6
zsNpk53nzUCa)qXxWOYjAYtLxq7f$sAFW)YYaMwA4^|^?6%*%NAHYyJuV7oWG??_~p
z%WS`WLGt;)h`+&CmM9(({)wJybN?Z(`^#g@<!{W9nZwKZB2QU!%8#a7L)z{lDQ5Z-
z#WdOzzp4j?{l-dG9$$-}wGN{}Oq5pA?5OATDjl)~o*yzY^C&F3N~Or9T3){%;C{?S
zijdSjfZu<`Gyi%6{MWb2|HeBzR`e0&R}?v|4o%eNPKjS(OPO$954>V$fs8jM&u4hW
ze{{A%+Gaaasf@6~z!W2te*M|;^Q#2jX_fe%Zi(#~PQTwS;zDwnZ;X-68^wd;MyPT*
zgcTk7o!CDF-%nl&-!owdCZs1#E*`>j^dHXqRPf!kO49>Dcqy#^B0?2kiU1)ggeitV
z%oFY)5V0+hB?#1A#Th$Jd%elo9Gn2{7M-iT`vgsVU3n9X@V@ep@8+leWcyFblN>u-
zi%!BeCf*ybDd&{c*6Lgp6ZDrcs30^DL|Bnr?GU(~NHU%NG})vwK*Ffxu9Q^nb7Gou
z-1ficqGWs-0nxs)f_6hr)@~;%GS{zS9gYEnGP+`fX>)>s#`N(W+jcGw!i}-%F8kB1
zqLV5zet7dk#CXAQq<zw<s%-1FmrPoD<##F;zh)zpC<}c^kKVI;hqD-n7x=mn8B7m#
z&j0w_UCJQucQb(hZiqDY6&qU_hG4bW*Ym*{WBlEkeJ;1K{nV@Ur>F)?e!cCPFyrnl
z{l-6&;;c$>kX3OhjKL8vmJ!?U7xLPhqZx^|PqMIFHI>4oxQMmRWs1JlvbgtbUQ5hs
z#4WYKA@oeGpSiQR7Lvj=xChEvxl+|gc4!|xB00QR;f9RoBk#KSQc_z$O^sYyFnhDF
zOEu_JCrD}TE=GFlg&0R$CljmCDSODirK~~x_L~Ql?p_30cLao+i6`!xCcer`mxx?&
z!adieN0*iuEqvO!<BW-vUT5#ZU3}C>#y2e}Qc2dI<ak2u!XncQjEr@9USs6ms+#zh
z{F+prwpP|w-+zpnQQco#@DQh?d!La(UohF2jvBP}yxDf6WdMPA=Fg@H(End=mHWf%
zahp~-eG<h3BoijWy0@{V#~kUo>7p~p6Q4$DUG9^37oXu6hal`D+z~D-wOy1l2p|mY
zBsY5iZ^OFKyi2no)+ryZ%vsTsI!lgt=BJ86YQ>bSc)0NYk)oUf#Yg1bjIWL8vH$0T
z?N7P*h@L<$bTjg=ujv1EODbHERyZlEm)4A(sFO3rYvFt;nchZ36_XtM)lnz*GU5l}
z4(j@SKR+s`1TE|qPQROhF3070o)7${+WF5SO;t6qooLcsZZ5vRZP2qy=s9X#QOdgF
zTGvwRy`Fcb5HzHGSq4|Z22*2=yHIz^T|(BFLG8!N<KAsnIt#9!EaAo{=K|{8*D=Zi
z(}P^wjbfdgDA(E!>f6~UFZdS`*eodBkyXnw;>=>nw0nt2ltIZHXK1s2mQCQD2zPmW
zqiII3DW}t3%MAJ3Q#VKWh33nK|6c9?535I!ykY@uh^(wEf6`iRPAVXQif}0I(G20V
z5{;%E5&eCa5D1++6KqEV0|U{~(My9l1mv7XfZ?r8<>fm!=Q<314(Fj@v1xz0xl9l3
z-#ip(fVhB0uSrAuQ{%Q_P|<J23-B+Q(1iQliH~n=8MK~O_e7q}ZMt8-ejS@mGk<tt
zJ2xC4JG_DjXj7rH@!p@x;cwjPVo``^r3+xyDQ6O>+9dK52sPr>p~C0qt2mR%-K9{P
zkAs@+^E^!6wa%7zf<Ere+qdK+(SJPk`k=6%7|e7TTV7tCPR{fc3t*WxP74(PiQL@@
zzMMmr%uk9@#uSMK!`8#Du3YKoK{*)FJ=u?v=!>wGb9f5c=$Pe-4Y$qTFvw3{spRXw
z>}}V0>?I!5qBhy+G~e|&UsLStU=CUm>u%_|&VIKL5J}%t6MepC8@r8XdJQvel5Hr3
zwZrksl`C;^amFbZU%m*UOB_Y%)lc?d`48`QJ$U$V%Uu+mphkw=b>>`b*yW;a1E8!k
zxo}XJmuPFp@4AZ`pOVNzv!?iZr;solV9Y1d9>W_WkImI$k76puZA)24?FU;g`8OMt
zjYAW(!_L6%-_dgELXjs&#o$swQ4xNqQPm_HBr*qq4o6su3FR^+z#KR@I1&fn{f%1!
zp;L0%nAW#rWn-KAWZW{8s}W9TQw@z&)xljB7IS2TXidDj57FlPz|;&oJhnmA9^<uT
z{ZpH>!m!FcbsO<e=ceyRP1kxHBPuf2td2rZ6x{>uhRKV75mSU+x}XuB4Q7O%44Td-
z(qF0_=j6gE9&2jyKN(<<eqe9E4tObD>H}i!t@-ACTl2Z3k}#&bP!U{Y0*x60ONOUv
zuj&aXu14oL#-TKye0n}Fw*QUD(+zrtfT;H}L2?r22-QoM@hlih`t5%eHxhd8#&fT%
zqP}j|x0!5?3<dCa1@lHek|fvTA-#$JC3k`ELy`&S7x_*A6)3q8qnrFmLz*3nbeEr+
zvHq<T%Ao#9ef+;1T&Tscib`jyJAH{FWohXurRs4R^n9zNH&v!^p`0I6Ow!-l>~PwY
zQCe3!{&qlszN!>mmR@Y{Turv3vy#CACX`|>pkb_3xjXDQK01#p)J?H>myWLB+3kFB
zT9`T(SXiJ$G6<H-l4t%SSM$G-9{UM*ygH@UzdmW<kPWMo+t5KECX%ZOJlA*RyJsWG
ztCB>#Q1vG7rsolea-mJyosNj1>0b)?+~R2wZE#4+k#@&v+2g$%2_X1~@Yy>C8y7Ne
zyY}MXR!@{0hbNCTyG?}$81EEeOIn&nP|YkCZWpt)aE4?AxDXM)GR0P)Ff>n9k=P*U
zy~m~yFcxy0&KQ7stlt?&bFI#h$<ks;UqQpMBl1S2Z-=L7`<kww_tBC*zUq8M0lv?n
zZ%OZV{jhZ*g#8NtkAo!}>Jj5B5ps-gzdBMRV3)|spgL%gmF#cxr+Eucb(W*(%oi4Q
zuc5Lo8d^5+1Nbk??Q+0gu`qr|9MR1cSas2R)+bDyhD+RGQNOZY5$-2lOcJ#sSu&B;
zTT|ioAz~~U>xE`O`sl@$gQ9bZTn)1EH{YpUa=Z)@r!Qe3dI)!pRIguuwNBPW&7GCZ
z?m2n^5&QQz+BA;oIF|Eh6R__qFLi3L3FcX2U%VvCOw=~#m+nS~K;cGbgb8}tR!K-=
zdsM8hQdHgF2$>DWKkj#?>ag-Yx!b}?)xqrIM*B-Jg79+??-9v-ugv|?5%-Z1_xh{v
z-lgQ)-RSvT%G$yyO8bNDyCD0jVEa*b#;w$a7K3GH4h7t<5ie53iyVS2fWZPQwo40K
ze<4!{zG`Y=syoQOK3M;l;&QH<GS||w?PaG~eUEGBy^Norc3l@TeoM)KdfIf?d)PH`
zdAVa)%ts|f&1!Dq?x+Bt{q7!?B}-bTV&A}q)tjUF+{9|uBUR!IAA;dR29ut<5pyQ0
zmA}eV{Tt8BknwTmLIo+N<P|(QZYUzYvc$a~LLGIa3JdJTd-fC^`bpmp9CnLy=$nMO
zq*}W8RI)1N(9N+_G=p7EXICi`w)9QHi7(sr8B8yu<kUWvsoli%$B-SVjLznHP0!K&
zvX+lXx<E7kCbslc^KB`@&+)QkVGK8}T*15$qPO6paegUS+klfCcSxlBVJua|@60_u
zU-L1es7PL<fAh`J=c5-&3r5lxsdthyu>Ee1u=N>nl83_@(su)iRO|aOEaQu;&NW+p
zT{uTihyb^UqG~LC)A`iuSH1ZDPfk-!YdVzk=br*<+KuaajWrx*l4iehmIuEfL|BD9
zN~KI)6r3@u{c7D`7cV&27~T7Yb6doG%5mr5@Xp}RooxzV=(JC440U$`2qTQuJ;bRe
zw-6}%@;d17tsDq<Hs6oY3bH)3tT?+yiI}*E=}qOn{%9O0r;iH%GEOJ~70#?G2f~U5
z(_5GEZkO(ST!ov2%PT`hY{d6(ggbWB9h?Ct4ie8;)`EUxy!aJGVMi_V<!b@%?%vJ1
z-8}}q%o~VAtmik{Uy39L^9QCNefH*(Z5s$0Fj3dFJX=+Fh=_h~*ZGoiEatii4N0Rq
zutYj^v`Lu>c#SceYkyzdt(@=vkXnxKbhoH-<Sys4i-;p5z4OU#NxA?LXPpj{#Xz_V
zZ>HX#6O*`f>K2s|Z74iwNRCtmr+L<;a%BReqZX;XbCTa?xycTFCKp32$|)-maV&PG
z2(h^=XSlKPwvhh4bpd#!QBrib+Y;vrUHpf%Pepedx-nj4OT*!J{6%vQ@vli&-wFz&
z99`DONAQyP(Im2dFK*~u^nR~~uoA`;<BfZ(Md(zQ9}$O;ECWcF8xYFAIrU70u2sS$
z_Nkh~-_FMKK;P*KoymeIP}VOjB+*r3T~EqTO-)ThBa^91=m^1kkM7=`d#wvtk`8bD
z4WhI9Gw)lPo1;n9su`c-SY{0qJ<`@XB0qP9POd73{V6y1y3rgrA;g4;>+0CocRWBQ
z=o=UmRbu=7-t3_5d&`yi@#Aa4+5O`jrKD`}a_{54I2Nt0Qc>@t9bg*1ridgEUdO#G
zakQ8@YVdpwC||WY=8!Gmm*JzPD<g#z@f1RfO-IWtF`c|NBdpS2fJ&?bs5al7U?ZrN
zKD{zevNC0`yBEeoazuJDT_~a3Wp`LBspnTnNC-<&H9a9VQwaC7MHHXS$Wtl-&z;3S
ziE~jC6q;iP;*B&^cCE>J_-?#;aou|D?5v6Om$4Glrqey$pcXFU&jD4_o(lQax6+1e
zXBxeM>x+hdboQyBTm)6k^Zj*7hkCXAnhU#m47q!iikzMo%Y{6)vO4B!R!h6tYko$*
zyol9RT5mrk+%$9EE3!c-eYCT*y&!RZn3rFacYZIePr&!=1Xv^B{A%@z92yS;38UlU
z09o9T9{@@jSU#lmm&aQAe9!k*gJ?V-#8}^Q=to0U^#9x>ycqyss{Z`!xZ$~}o3K@2
zfAE!xDy#`RVa*FDimNy{hTglwIjNp^1qAj1-Te}@2kk=a`I#qQYjSdO45LcmUV=HY
z!MH+4Lqp>V-SPfL6OggP?P{5FKrB9m8^#ZkvFq&V@dg>7Jdb~i)EiC^ghvUuCu@1x
zI&}f~d-oo`Q7b+LD}u{Hwllxn8qZK~kjN#0pa5|MHC!lJ-PBo+z>IstGyokRVAug0
z0~S`JovN_W8ZFWniXg(2vSVgr!%LEulWT`y8=eF{7KolDcn(TRJq`QQ-JD-X3U#%L
z^l~3p;H5^MHDl3w0fPnfRkAW;%Xgg*C|k0(N^*fqrWK3&*e~{|@H^7Zc5*;v=_=LJ
z(U~Z->?<f>bvbEBPyZe+FzpUkL5+VOzAee2@m+HAN&kJ?sb_;q?={G|&A$BXj7jtO
z^=@Zp2SV9?x?YHfhetx<ya=M>WUJc{L;~Nl?KGXoli%0lpyB9<q;;#CQW4u7vP|>N
z9tJs#2?dNh1lxn?ET*pI;e2N&C->ADK&7%|`oy@G$=GzBtyPR(mmVq7XOWK{UC}Q!
z>x`n4(T+duHayEvZZ_id0%ZiKXVz}}iE?Hho*2TdVJ$;__-6KmmCQ)-+N0(C3iQ)B
zU7xDEA3=4Pc79?BjQmus%k!lV3EbS=NBbMa!+IEqS$$YxTHljR6lwmjCv4TAN;5Mv
zQP^tl*(#(|upI@d;Gp4R=4JmM5hClT`k^mUuo?0g)4aGPBz!v}sL_Dez+f6;@fr{;
zn@n3MC28r;!9<Ml?D`YHtHzrHu^1iwbjkqnqMFzv^sI8yo|4O?73fPeoro;6U^8s=
zIz8IuD#Aq|c&`a9+VM}qq6|+0$4@TdBkEd@fq1R)CdmGq{r+0jOlnfnWQl19P}}B~
zmgm6pzjibS(lGdo?hXza`!nGb&>AYneZ;c+tt;9;-9^0MFWkFl(UU<#MEM-D6>-~2
zag#VBBV+q8G-to)u}=9$^_({v_H|fEU1)uzAL#t~!o|6WK-k^GKLX1B(5n$e4ZJ<<
zqKzgA&Bs8#SkPlNP^=gf@9!)Pf_NjB$<=NShczigM|%RU5OMSHU=h$Hr>35Pa?x6P
z2l2wyd}-hVrGP!IO~J>HOw7#vW=C)QedfF3njmB#8um73d=GyA*j*XTx9q?|ya;@1
zp3{@x#9errnCr^75)469+(g+|X*W6Td+x1tHB|Ij8Qf@+eQl>LKan&v9`_?@8caLz
zlcQ-6>2!)p;k*9h%XUSp<@@mxgb^b98h*{+v*cqUXvD0`#)Vy1nXo?_o*#`u1M72o
zq?9bUiT<nE9}68VKZ0mS_$yb(hQCPpcHT~?hsOjm1cZW5(|IUIqdV*@T}~GSB4m8}
zlu^y6;TiGvB%d5M-M9XhJvT$GF++a?*X#rP@TbaMkg(a<*w$W=8K`m$9l_)K_46nC
zd{tAflb~MEc>H|BrdydT?k!wiqFZIp_L0X4h0aN^RItOBQ14q{aoQuOLvh+KAObee
zeazPEDl<OUpA4Rh@vd6TS50XXvpWr-l|tu*UcPsVXxZ)x9Y|RQAc^X`O;tEAD5v%1
z>;;dF>ERg|FN11Sw~+cA3X$WFS7ffsLrYXKZv`9-)YRVeSUg|O%~^Fl7NWpFSkb$D
z`RF{kz~x&OO*mWkr2OsF0Y(n+_8<WCe)$3d*F9I)J*ai;pqVmjWCx@UiJO=@%i?>=
zN5l2jIn6)Z%?r$w3-??|ant7i%u)4sso{LX_q@)eEkxI$q2j$90#SP<LGO;96LRo4
zl2+pEeGtvj)P_$L^sKNzuv}h;^9jIIz-tD$0#F3^c1H~jL0otr4$_3Rl<^z?vlta|
zXnL7BI4sGtv$HLJetCo5maX;d?ChGB-yplW!Dp$$5eUL%sCWmUwMpyJJYVIK{sOXA
z*7Q`phx3|DMK~Lw&>~blA?qQg-39%i7RR}80U9AE1{0OYj+m$@N$e7x!iu^3*C;_^
zuCX2Gmi_|E?a<Sgaf>3{qAo>A{)&wy`JE32nYr}?))gZTkzAb?8ewuaS&7419I-|p
zPu&p+T1~VU!2i?Vh>I)o%F1_kBi@^mx0`Vf+lWR%!7|~ma@9|4;=U>q@PL*k(3n`Z
zM0;6p^BaS_HJ+U2LYW7SsDqPWVl_RvN@~gE8h})W{J6J@*R%qTMny_POZ!(w>n-G?
z%TtVWiG&s%s(%DRKGI>0UtL`VK*6e4{Tq_JAm|ZO_5_qqi^HAL`WBuF_V&9&uFF3s
zdfW3WFYm4$SqlSp^eD>+n%w}+Xa4HUdyr0!wB@!b5<|B=J*1wFvfQGhYhKw6$snIt
z+;qREHq>B2AEGI($DUxq8Ni>^V-auOVBnzbCaabQGX9B=T|V$K#%yqFiY^sMwPlqN
z>x!<@)#AE)G)PSwl-QzD#^SzvR;~HAs6&Mge56Gf#}6cWXi;vQ_(1PA#Bz&^tAEWk
zNcPj{3JIpvb|NxfNl|ggHPP<j&vZiwh$8(uxkwuQNhc67uSvH<Nco%`pi*DGdW9~D
zP-Fn~)Vgi>M6`owyn9?zTB^Rip(3khhQv=QIHHw5^<YLj57uXX*tVDt8ImI7<tfM3
zjeJ;Sa*zFcgbW6bx(6tD(=JW^fkWdc|6xX0A>?dyA@|&`jFob8D!6*2m4>lf@ggma
zeDrcLww>{OxX2s-MxS(nEKo2ZSAqn4&%|VQP(?xs`-Xu1H;`jnV82On;H?`VA3chV
zj;5C%fE*3TAYsO9y9aN#9`>vBMsJEpmCr2%V)6D?jlJu}w3yS+FNukX9W>9w77h~(
zJegN>j1h>}WN)e#wF-51f5xc%T3FZ^E183D>gn#Dtada7Xkw~Ig+Scn0G#rF7OCZp
z1xy6Okq*ix=)e3Xt*^(+EPX-lF0<&-Y4kcg+Fdc#yMchz;DnmAtsKtR0#67zZXncH
zSmj3Zg?9M3lI>(A8_>_6K5>>i9`CKmWo|+Z5Z!maMKJXsnis%jK|z6v_~FmS<Mrd?
zne!rK3CI##EL{o1d^R?sr5UHKRd$n=0FghnuT55wk&y5meS_KtNinX|ERoNKl<w!0
z+f2iJyly;)VFVA!#UV1)*c+C;*QE_OimB3;hfcq1qe4|Q*F`-~bgP00xkdXo?OiQ0
zX<tF21*KNM&Q;%Qv(4}3$<CmP+c(Gi*4E1)L_=lacmMr8RqXk7Y0Uy{HFb4$uzZx5
z-k1KeKjYi<WH1|^$=S(vpM!}VEHQXS1guaucb~=K2bn^;sdL+~U!J{C&N$fB+bgUg
zh5b&))xzQs+0gy+t>;$j5~I&?-SJp4vkLa%=f}>zOPv!z25g}x&10qJ-P?T<GgS^V
z=%Nl_iiKsR5AYKlpN6MTBj$+)BLJg9N!b!5e0_X;&OX0_4rpbx*cS{XfYeX!(|SJx
zZC0s(2c8_*M$QHlIPh+8x0-Rs{RR7KWGg;73q?paP0=khjQ<!sC=f@r&y(X`J#=|1
zceJLi(InQ&c;cEJMcmjdqp?vv7Xdbmg`Yn~fB$h2a1XjNo*ZtY%Mt{kmZm1VZsjUK
zfixj!3-B%sWvVXprAZk5fk^8c1w?7y9Y36^xcZI0uhIAX3`Eo<-Y4oVE_{94M2yO<
zmrOC}m(6&uXV`ZzA7=lK@w_laBgRp)SDD3RkA$_6!9LWm-ZB(P9*t!N!F)Fm?N49|
zf&RzFfik>(g6R08z&*7`v=AE(0Nr5SrekaF(bQg(-Hh}-nFB<soGQBHS_jwM#BZ*H
zSVI{OsN1VNyczIq{+W16e4d+7I%S0MFEdN~<%#(YdykmGq}Z)%Q?S<H?9qWxhqC0d
z4S?LKn+<)W#N?ImGv{u$suQ4om2;edMZd;znMCa;+tCdk3qt8s&*j{oKYzZaMo7ke
zapbUc)z*G|sxC+26a32W*NN6>v7y9n>jp6Ff4Uc-Gdl*L{7^vw4dURd;?n+nB`149
z6oH{;aOoK-De1)89$0;p)5M{Wua~k?uIN)`G=7ni@H+Snl=;v3`C$V#et_LPJY`GZ
z??Nn$pueUa`hj7}<yXewKoEw0Zo+=~LyHZ%QThJI%jGq3?;~DwIe=Yr?cuO3Y0nSt
z+jRra>Q4k}e{1F!sYq=g9_2JTwlEMkmH%WNN59(xUPna@4Y^Omk?QUs@Kkkvv_3+3
zLR%a#r<_*auq{l4tsU>x>MrJ)aFLJ=Xi&kqY=>dWlOO)=g-JVXmv4Gvd%k#j((GMP
zRU%73jQ${wfyd)DyUml-Ba#d0c&jcK3E5E&FKjE4MhbQ!8WnoK4%AV-urdy%Nv8>X
zBfM}u^?uhy>ObYnztd`HO;hsuvxkI{mmshEU|=DtE{?5SPM;EDeoeM6_5hn1AM-B5
z6J`D)=6gc6{TItc@~yvBR5d*ve)KvzqD>O_<|4)fHDkxC=?dpB7+pW`E?c<;-X6Pu
z=k~`cOS2D@H}nRRoq&k7a&r&#s56y&#e}Ew?As5;p*kunm$3{is~lWhf4m0Q3;#)9
z{f7(u3)39?-R=CVW-{N&j@X=|z%=Cq5y-4~)h$?ThV14Se9pd$Z|BX!;@RQgzeS{D
zxSd$7*}*~j?tt)|p6X$~)NxA|8Qp^cjSQwhmwdtxSLJIzWR>?{Np;{>?tCy<+UgE0
zOZv;`c!TG<tI1@BkYkuRZ*<%l6}sX7SG}>y){ogm<=aRvJXEWj#L0ap>(+B4N&_C$
z7e;aW>yj<HxhR~%0#Wflf0`wPCNLyRBQHAcRS+QXuIx3aD=U*6;YG!0>=a#^AkC{>
zY$&C|tzoC=<BVp~9G;}dz8j+QW~FM>(iln9`~51yQ8_5xQrCDV%9uRf{QTL;kgpz}
z(o$1w6q!n#z{J}4PC27+mYlx68Ey+FZi_61ycv?eK;OHc-Zg?*R;^ogS#zMk{0UxS
zpWl=3mm;gPs9HXz2x*@8_`Rl0eZ3J;JF<B8-uk2H^%epzn*jBs3>nn6Xt9;Im`S|(
zl@7MWM9LAX5^eCLRBjJk?cT<1p-PCW7<Oq)4l`(C3NdbyI-T8(3mpi0JRGh((z$(0
zX9IT*R|cDl95>{B@6)>-j<_x4%lvt`mt|UA?VgC^v>Uw^MDoRbp1=d@lpkn9Ym-u?
z=Wav2T<&hY;UHL>-vSxi%a;D_`%_vsYCmyuT<f!op|a8UqzF&WP;#jYJr-q6_NJ4@
zgqQ~L1a)F*foqT3x1$+ed6R{NIMMXI+70^{^ODwt>xtfq1Gzjmnh1)k-AJ!3lJm+1
zHL2-?h%#@heN|9})fs!)%NKkDW*(_v_$biVmu#BhSfSq0l)Z*2RiK-Cq_69XG((bN
zO6`-hGKC~F40E(tS6sgEa_oe>+cCMvA{6{B(!=E5qRS3ttR*hEc8+LQi7p{#!@~=R
zE>&^vO9;fD_R#-f$-(aQszz8{;FHf4t>kKE^#&u4ALqQ8BUUrgFG9(+nLG}Y3(v<y
zShY-eqLQ>ugG3zbH+@RYSVhdg)r%J(&B$%0LW_^E@JOHQWf{L((=gtAnXT8wbCbn?
zFX59?_10EGP0CfpgA13fNbJNGy>FePW*4?Z)1snO)zsb_i=`h4OEB)>4B7IRv99cT
z>WVhwHWvC4W6Y9`uv-O#guAGpKi*nepxb(>2V;W%gdFD|EEE4^OM{QXXJ^7R|8DhQ
z4#iq~ifC`O4eoQdpb8d#%7y)yC?bh3nn|%KnV^*yhHW0dd(m`d%8P*<SH$-U-95Ui
z8X7HKTjx5sm)oodtGo6vq>9BVPr}|WcC;UNUjI>jBxatvzz~cl^!w36h5IO(Z(JuQ
zF^=8s?Z?dAYQ!t`(Vr~d7=Lx?Go+s<I|#%dE&9V16HdX^l=OC{uua8HI7J+u>C;%|
z)~fFxPW==195x4IVq)~!6NUnuDWo$w<^ItV@=tH@zj;vEA14x!pB;4cAe90QD$;#+
z)KdafCcs2s8)iHfQZ)1Vj%pAH$?kh~vKvk{fB|E!5Z!s5QEkyfQBhJ~@713n8!5W=
zV4*h!00phf8>l=^{g+PxbR6%lK!b7_k78#ovdQ4AJR`&8OB5YQt^oM&vayBNJ^)x+
zV-;k7_fN))j)9rEbKWU8{L-H<Xxo7!1H_azaj-Y(P^?|zN4RygQtY-m#?^Jme+ey(
z<juETi#Mmn4^rPR{bKhDEW!-8Z(qygDt~889%zhv+26qB$r(FK2rHN3X=f>GK2u1i
z*;PB<2hlqIZ_LZtGHdQdlufQFPL&S3<(!q0E60x8R`S>RePT@lp*o2~Dt?~)tL%L?
z+#~RN6=|2yi-=4vJ<HSDh+_Z(W7HL0q6?5^fw(ugZO1q(%)t_N5euIoV9d4>K>kM-
z2dWN7uB3!OD*T|n1tM9lcpmc;pTTcc6dO5&CcIa94Cq`gh8vShKNu5UYC6U+c@mdq
z2`c_)9bVsr8<8DZOuLLWCECQYo5!c`!~fQ6l~U~-{-xJ~_L3u%Da5C#si){BEGtbU
zBR?6IB~r7aCzjmG80EG$!FT(1n;CbCxVI;8RRK|O@^9btcGVKPciSl)M7O{4KyOo_
zBCmpeF$O%+)NJQ&Gf94%7{$OmV$KuhHT8*9V$znM2X}7!P@3{~{H+q5t38J7n$h&i
znrg(z_?8J-q^3xjCsu=++On@Zd7v8c?b*`B-|{y2YQ&~QJ5(KQM(o1Y#wguFO&7yN
z=#s%}_!aOvKwkis3;~KM!n=TcD<;=~7`hz*_~%FMO3~ma6DtxbMwec@%`%ewIAdhz
z+Q=t*1nObA=?-hR@;mPN@6G<e?v#w%21Na62?Bd(s@t@OHCDwn1Y#$n*$2LugqV0}
znvIo}As{O|8!bg6UeiSUyD5|ccM3dJ=}h_7%O{_0@voM$V)-|t<eFHCq~o<le=Cix
zSiGJpIyQKVng-NC@iS)LgQN&W8sVJ(Tsx5e?$w%Blf#dBwaka!;Zqh2*q*PM(PDD6
zpuPNp@O{V0HCJSOgF&&uZH#inzk93xgO~q*I9dPa0#NJ*W{;B2oZgdJf}v@1^1wMk
zv>l2X8&d^))<jDwljy|s;lM6_UHfBcTr|LNZfF^qv@U9ZlGACn<=McNY9tUNA6hFZ
z0H>bNvo8yr{;LHuWH6rwMg6~aQ#map+1mVaGwjP@g6wDP1udnvUU5B~VV=h74OdZK
zBj_%DDxq9$S3;DvyK*y;@_JdKDd|Jce!4}ER4wA$mwD7BX26+td3%5m^wmBd+)OvR
z9x(Jaa#Z{(v%n2}9b-2~dzTM2UOT^Zwq71qOW%6OZ_dWW2IEg0@GzR#|Hh`M1Ao$~
zk1y`84|Jhz(8pYoLe)iC7y)n?f7>wRS|?T@yN&$!x?a=c^RXa=9@|g5N)-Gaoa7U>
zJ$S%7y_CGkk8J*X6+e^!Bg2dTvJLn@Sb_e0zDrNlc@Wek^OglTGY<7>(yy5&X59Hc
z6JzH(EAz#>NNfej4P%sF#S|;&DCHR!*^6ZJsU5bq>$WV$e2pzdu!h^w87$A{TRWv}
zhTQY4V8KowHU`t?|3^1w{q!Y-<lI}y7feobr)KXkFNV!%E!0Fm(7sck#zlUm<B2Xq
zITBhAK_h+61U3#ExJlOT_#uaQ?2=KtO@gM!OoE0<v*|rh=vp!Qs<CuW^!{eA;Vr<L
z#rbbWCHSlzCM(fI@4;ebq*+aGji4DamCr_VXAtOYE3jIFGyyc_J5FOirVyKvf-Bxu
z!+DRPWQIB&0WbN%%HIbHALLeNK`KzujzAf-nS^<f2x<|~D>9PBfwy6vh)I0>=#gNG
z@}IK|-Sc^NhV|~7FdV`QjO0W-)ZORc*b0amHS`5CJOb^HhU$gXWaEhV#JciSkXW`K
z9eZ$$+`l<p3xn%<0y6$FF(mdj3)e)|nG|=uYrv(OmeB{`i45kvl4=50nS@2t%+eBO
z9^|93>7p6l^;qoo^Pq=poU?)D*<rpV7e+4l^d{P1Jv=mYPs=stDHU0n$CEEVo*sCK
zs}b+AXWjT}!PCMio>shBqyv7at2V*ob+&)4M<wr4?zpE`X|p<3(SzUFU6h8;K!>us
zRM<YdRT9U)5wbZmGcybwfYtytoDr?){rvSSFS5v>VPeg->F`$)aMA|VPk#Wtu8jSU
z6=;@)>v1$go7Q+KYZ+_4X%AOYMFvwTYsY9^3ujmc)8^x{hs14Tm}2*j`LuP&cfg=#
zU2$C+$RAQxZBCFBq5oD+7JNg<4YM{twJA<ji$Gxk7Lk{g41q{wnc(t_H_uUg=k)t~
zGj!%-w*1mxe*b6z$S)kW4TJ$B1A~?79dOLVUe<a$D_E4Cp1x<li1J!9!UrnvRq*LF
zHQ2TJn;g4KSGNy`&Ut2E_bu?eQ)H!7!D7SlOJ@q{w+ic7UEYvX2S<1k63bu8d!NG^
z0~rsUL7NFY&0mgnQ-G}aS^R2)RB$b_C|?<XOi3&J?Bn!f-pRtM=_3#}m-S`0L3mz>
z(=zC<3(<B*bBoV6#9-zpmDcCL7$&Jawfb#k0&#_&et$9jSd~7NpAxeEC0_DOhcdA4
z0WQgKBk6GkSq}Jh^x23>N!zFFC)ft_wK#%WzzEb6lbGlUO6Q`!Y#AQ4Fmauem242&
z-ue1A)JKGB#PQ_JS+*M<8(qjOOn4-uq%?r?0&Fe+{!%o95@^jpPXfsWMt2ZE3)RfL
z^EC@ikT5Ay0W&VP&4=nh%WZ-*gC^mU`0bWwntW{@X@dlN3}ObT^J+-syZHE#(qN!+
z1V9Qq0Tm&~&Tb;aNgf81?kFWb($Nuin63vJ<l}wf%1S0JOwgQx1Vl5${siJoOiZuG
z51(e}U3&!4%?do+NMTZCg&MO#)nX)#Sgp+{JItnmXmWIV27m({{uvq4#BDHFcIg6!
z^lR2`a5V0DHteP~0(%MDRvhTFNU^=Kt@&<K%{??z*UizivpxkVh*rp{bz>iX-;2(I
zQ6mtV^Sr^=2OlEcT!Qa?zFxBc)U~0y`t>rdD&Xu-;HIBEc>)*<EdP*6fawuW>zmu8
z68F+(Yk^b2^M0Q>eDUQUzURj?V~;4qnG&AC(!rosH(+0M(1McJmi>!KD*%>L57_1X
zZ_a(_@HpL{>BzRYz8fNmE%a<r)zQ(hU^)ExX05C*#57Q`pxfAetLD0OYYuc9krqx^
zTGqRFVK5LL76sr|(1765K?Mp0X7^K?02`@-l+?>%Yc>#HSXe^GM>f0SSUghCL39E3
zmkT_7Z~?Ii2~FF*B7Yg3!Vj8*``P~0!@|OH{42?!sCqW!Q;wli=k3e>#(e<sz+JjG
zVPmQRZ%Kk@35LZ}pL1{yh#xJ2(P_qWr4UT3Ks}zpWZ}}%QgQuONZ*1kWX2B(20FeM
z?K-1x!$uZ&`uTuzCL_{G9>jbRw{=@iauCfE;^X5J63F}$Gxe^)zQ20)svi3&<j<);
zHSq%SA;k=XYuI!SV0ZwV!R>DD%Ecsl2nmk*-=859!C2QW%(kpA_Gbdy{Twz{B$d#>
z$}ZX{3dTqZKI<VcLxLWD9w0XL<7rk_*2i0Pz6UL|CYr7=3kKn9vk^fhv<2P=Gja=<
zrh`qW78DH+|2#<I1lk6;vrjoWSy}L1th$w7O-fpSnW+Rsg|gRvOi#ZlB!p}bf!^WT
zEjh4+M*<`H^(_}%h%WH;kcU9c94|I(g7LstVhyFGTd>TaD!*2$=zr}bKR7seBhUks
zLj?r|EIbP0B=i!nl7_aK94_Y<<I{+K*)EyH;^-nVhAIX62|yK#OC2H)U5g8*@D#*4
zz<&>pR}gyjJ{e)q{q)0O|9Bn?1nicmgxv%$btL(tZcQ8`a=G~V<>lo*@#8Q;Xd14&
z%8ALURoadZH2I!`D%A*5Jr+I{%3N2j5L&GYTV#AbMAPwl9T4?kEZ$$I9`XIl2xC!E
z5h*69P3;yxb%=WO)gwE$A%yi^h947%{zkNLmT<h~^_3li@k4d>0l3rRnqP@%-7i{!
zJ=n21?^hAFLIQgVr}1a7@h)y3!bHJW{LTnNi1fbFf}7Y2i|IZ;W}N%PAl@F`zbgZ~
zC)D_D5UtOc6>s}i;>Gazl!OEUSZk4u`Z93WufDYnG$!RR0RP0YQ;l+p@G$r{G}-sG
zSi3L#H#<LLVr5<ZR<jM7r6qsTO5`N?3T!9JpMVYyuM(<D_Ag<Jj+dH4{%H6SNNI+Q
z80TP8E5e4<a<E#~hbeUc4xh5JGTpKP;SA(8Gh}aXuK_qAU=R4huqA|hrBey->rpn(
z)CFf}XYep|fQKWrEn}TOCO$r!=4*69LPAW8GfZJMg2=vnUTCaRoSUn*|K?`Ev!#J~
zpvz5dyBc=}6cSjG84*80YW;Z5n(tqOA9M>O>p)8T1BhYJ;9!We{fM&3_h>K;s^U#~
z^xCc%dS%7|4-MXhZzd(rmWMixI6oFYMauy(QE+y1XwO5*IL<)f@fkNy0PC^X>Jre~
zJanxqyFPp4mdkOL69r5OkK*US2=Y-&x$#$`jfr!ZdMwLl7|(;%{2=bxLG9L>ZuFR!
z<wA!efy4<Xof$Cx><Rg<V^@3MH&Ag_fJO}2bD|DrS?!M*MPQH*TqH7UQ=oG~mS5CX
zA!5_7E#IqbGYRhSk4a*pNg!sCJ{~fFIq3AMkWqAXvZDX%a10^s-rD3opsp4DOOkrv
zDoX-!JQuhJBIy&Kb5O!ySotP<+sb+zN*PLNPF~(}iraL}a$Y`UjiK#`+YAio#|_Oe
z>|;Q7z)Yw|-Yd+_MV6Wc&ulcE9j!Pvn7q&#%Lw9MgKRGFP7ysaD*UX>VWtt(6kkaD
zAg`mxYr$XxlecJZKscp<r)cU*QC$&CXA`&C6gTYUr1=cgxxT0b=Gkt7L`%=OM7yYd
zt7{_O&(E(jh7oP|4u}HNKm-NfGW5S_mjOtOG11Y<$~kQ&HISy^Ef-JoE?en>Vb3vp
z?}2y3+z?R|7j9;1T-6@EFT|r=bLTjQ)hjm7*F1PB90h<RVQK)qCqca+mdypXnxiqo
zj<j;|y^?~WVnJS>byu4)IABIW21Cy~+w%9p3Y|R(z?dj>Z}^Q}PITG-J>a;<kI~EW
zP+oosTucuiK3u!f2D$#~u9r5Lc_70;9}ZB@80iWoFnHCH;-dn;>j|LemcShN*Ab?F
c%1j-f2l;X$u(>ebf*OL5msP%3ByHsPZ*JXc>Hq)$

literal 16253
zcmd732T)aQvn6~8Dj?tiB#1~*KqQMK$w5GZAUT6%1SIDyQHe?p5(HE-k~4@12nPft
zNfyaTGLp|Ue)pUC?wvbT|I9yAGpEXzdN+G-&VKgO-K*DH{qnJ*3;`Y`9s+?NkbU?-
z1%bdgIRA%(2}kb8Su(*7T>FRGP6)(h;`4tP-zR-f5C~d?>;rK%_c!a49^ND_XQG?E
zRJE~HPp#jQvpu^LFG=*6&<ukV+kyP*1$`nZZ4QmNy@Re^DYZe;whJzrQufNoSCYzP
z7fdhA5hGv8<IOWa#*6#T`z*!F+}AH+m|J%Fwe$PK-97K!=(*U=Zm*bbmH{Pmk=BPX
z4-g1{gTk>32!z;<L?0I7F`@VsgxGy-VFY5{eV$y~z<?bmU>kM8ik^}2Vntj+LcR)a
z-}71if}wm@(w>LGnbo5<)X_4B^O!sVflY;l3h{FOggQkDzcXb<2;d&bT?8Ud@*g**
zY0J;g@8csNGA$pCq81J_VOR>xjMfOC?M@SP`}2F!iX1cUN0ZO-WsZi;9=>Uz<F)cE
zrhI+E2gCsexVF{D2V1kQ^^G+(Kjt0NguV86mj^^LdKY(Y2zyqPmBk=`+4F~mh2hY-
z)}Njn?!bR1YMnkPWw775gL}U>(a_hUBkD$2@x==Iwzf8M4g(>N?HpR1;Nak)p&^Uj
zbcP^!xO9F83F_9Bxz1RTjA*3e?vk>^*HV)XDnc1?aq+J{`(~(w#{DsCCE-0aZ|{?t
z*3fwU>y`!+@*Ishn>SgQnRSNHA?%6PU2%6j8+}^>E-)}K<mKh{ilV)xq@+fQ^=~Vs
zui5h_u;~t0J<|<s-5Sy~+*=-4{PY}ayz1G&vNK2XN!h^i$x1<qv5CpRzyOxHp#4uV
zaU}jlNmth@=GAvuK|w+O3G#{Tc`7e;nhqCIX?zdf*3`H!bSGC&`yOAS@&4#A+n$q?
zb8>P5fA$0xV%mGB#~%Z03Vx&udr{(tBpP@w59`{c@je@aQ}k3+$}Ib{S^_WK!=DxH
zee>qc+?=U6GJ#c_%YE}_QW8bwu%ZW52dr71zSKRH8UCYT^MUt|PLKAE2Qtp$Qc{iv
zlthUn#;Wbm)y=2t^&3s~Ga=W1H|%^Xx9Dv`pB<ZYL=m_f8yHN@MvEG#sBoFIM~L_w
z<thtSJTA~I@!e?rLlH6xZ)$?ecA^IUd<V7Hnj!j#yu+cSaX+PjoaX1-P-?kHk9>A|
z_;z-71l@kGp&z@t?jP<f!rS$obQyW~{ylRkW#(0EpYnp-+=p^<H-nb;JZl+vc@vN-
z8X5^PF%md0!^55UsdxrZXzSQ%#miyQ(SJM}@ka33-q7M~pSgbu4MTo*prWH^XaBbB
zJgi{2^EL*34bCcgn&&Y0_4V?=2c;y=w=&VxJ$fQc1U|bx?1F;FA2NK2r5s=>&elCV
zJW~1W$gqtj>s&_koFa#;;*V+O8hs_c!lJ)XTs+!a6?9q2k(lfYSl&DO_4jjf;*EAK
zA7w>TI&c1LdGm&<U`Xd5A1y5=rH?MF8lkPFcGFG!>*JTYoxZ)fU(&EGS8Y3qE&j=j
znJG~C$b2MQCK}eaal4BJ_8ETblCwZ<Yol(p?fJ@VwUL*Xm9<CRDCoBoa9P=1NVR+a
z{(V=yerIPV`eacwQhL@TTA0l7o2;xXS-4bx{PTpF<qsL@e0J&mmfc-l-z~Y5lanbU
zLy+6u_j=(1^p<BM38MmMvVz;R&!1my;g1*xGQ|VuZW}bXZ|#j3(%dswuqSU&{NQ!G
zUOhh!r+)41RCbKa)V|y0=R!_KR+V3BtC%iS>-FdN>svDHS|4eJg?)a0dQRXm3}NrO
z_G4?a^$N<3&*sNZzk{BSMy=iEPUUge2^^iAuypY7@YM1iu+jLPY^_&ME8-->{_Re_
zcm1GHHkR>GhKPQ_5V3@uoE)#skH<TY&CShA3>s6i2XixrtnS^rH&mQ;%_zEQ)Xdb>
z=EujGU~+?{AEhSfv(xJbtg)SOZ{FxRRX;bp(kHjOIs#jADwckEtAobwpuDWSTzh|H
zY7AnkKU;=R$))etlW6L!EBi}yWx3PY9nq!#OmK?9B1d1oJTUtfVWzE7GIC6lMNNu-
zF|cFyWpQjM4<$h>C-eT-%BKU;$rqkh>d3}qt#)6Gnyt_&LL2QVw^(x5>9LzhPrg<X
z$xOl%-`Mkvm%W?5ct0w<_*o3nm^bx;_>1DQ`Ap4H^ER@gf#odrC0sgIl(c>iRmY*X
zh*?>{k@!TP3)VlA_rE6I|C-VN!@*h{?DJHa;jxgMr}|-R-4*Q`+Nyl-jQ|3X%P^72
z!OYCe%KGN6iFC8?sYZ#xI=b)sJNdv8d0Yen?=s&KL}R_nsx;2>Smi3bPpdHwy_zT0
z<Fm7~t#3pTh@d+Y;yw8V1@qxbl#$ZtnJqFl9cNhE^R&goX4kOQzCDk^rGsSF)76FN
zR7;SUl$86U>?Xvm#yW9|M9D}&rDgE6S$2Q!+^?HADfzIW=yV5US%3V&h{LyJy*&w$
zs+W7TIn%09z;owLr(f5HbWgdawdCH#dgo=8B-AsUs99rU<1Q46R7(RJQkc4?rp$;|
z;Kjh_W}_=BR%T?k4TUB%xH6a|UUZrgOWmytIT^C#W{l0t%92=@pzN@!)X>miCTaPb
z|L%OGtqi)p3z44Td$bI@bAMxcy4J~}b_>1%WOncM%8_qrf_z^4YckXxLCCC;)p_Bj
z?QKVfOBdA7bY9NH(%+Fo*)oJj6}!nCMUDk?rx=^b%!Usz&0O@UcU@OeRXxA)j~_o4
z)4igQ=jqU6XCac}ZUA!I5B<l_u?q{~ucKDO@9yDIYfMi4^5x4rEG()|w24eBb=F<I
zoLwd<BBc||u%b$yiL^}iynFYqi*?L~H$00%Kanl`dvWT)fqQDtd##a3<K|1bn70)$
z5RVmu<#AP_MJo)e%}g3&?`A!uSN?@Ljrn}mIV*SXSV>;t*&-G1%`z!WvDN<Xfd<tx
zg2fd%XUh!+pUZOgNqw({Y&pCs-Ivu(r*8DozW<ty%fC;gBi>%BF?Orta?L>S@8^M7
zV<yJN6NI7YTJso<>odd{nLJOI#WzsjNL9bDq$$+<l)p-#%c<P4Gu76x<`%GaqYB6E
z@#QWx<-`j)iY_<jNcbqw*vEK(6Qe-AVmvr^CHs<pLgBkhjeYFpbj@v>mwd@oR2pvx
z4q454Ffz7Ze<s?FiFiQf8*u@VNfkhi@OQ-VL?F88mM{=Zp_!NdA3C^YMUG&@oZX^m
zGSERg$2@4D&gv%)*|Ia&PB5a#raj;jK)Zj)8Ih?{L|9^C3nL|7d~uj6s2dSYV_MPl
zS*~-zp0W0Etk7X=Ei*_g82&=}=z%r1KO^8cx*%lKj_aD*hFSzIQo_4dQ?61{&xBi!
zb+vLlwLl1ggLU+uEyqlU7Q&LOj)u>=+QxEwpfQL7OUJRqnV^}aJ74EEfI{xey)HgN
zBfEueT<cT%ddpIaJ(5^sMH<#-X77IbIbIBHu)5h`kLY8ikFtBnE#Dyhi(l3b$NSTh
zI3a_)eQozX?xpPi`UaStCpU$9%ZSbbjI<MGYy6#l-;cT#WT~_A@_?%2E<Q<_P`X{7
zz7@VCPQN8LhDCcBE2^w8jHA*xJVCUMZC{d<xOd0J=ete-$_9rrlJS=|Q4E*7va&ou
z9HYn9+To-9m6Uq?){m?#*n3;G4_-4wB$z$)X}{ebE&EZrMh=0a!tf%|TLeu;vf)@O
z+%j&5wX9uL6q4_KITVj0sCUtNJ5`h1;OYf&ifblkfm?rOi0?<;OIC=FkH;Ws*3_9Y
zD5zy?*W0ieU#&1G;VXN0rii4F#k=p9(K`}>A_x>3jvQHyx2i_fn=lU-{>X9>NMj0P
z+nKDz&x=yhk^h8)d0W;{pE=d>%(%!k<aLh!6-pxdU$<hS^87f~b+UbxrFF2ZU!jjb
zP^eG4r|k+zIP<u9q(?*0lVCNgC5+}fI**;1$CeFYh^76mYM*)*xo7t>IvVgehF~y?
zg*j1E%{P6t5Bt)!d0|yR`Tr|BUoFywBr>W_xPYbLL>J#Rv}0gb{FvC|thcfQAH!dR
zP`rdIm3Gak?o_uTU~cE|f%2{$D``+Hb#G-zCaUOy>_FI_A{L?uFTlV|I5H%9d??JW
zV92U^9P48IsiFGbI5~&U1iRGe3c~;5zwC?uec^gJT1<7q@mCz}sc)j0w#36v>ZQ;h
zD06p34_VlH%DvYhX~DzDycj^8@+&y!LGzp2o{i~76dwkzagP2Nu&~Ifko~oaPSS62
zB4R@#d_yxs-yZ~y&iRd4a+k4MFJk_g^axot<Ek2XqIpB?5ss&&ev(|O?QGrAhgf>)
zO<C_g&8xlTK}$X&&G)&AMWZ52Um{+IWQJ-;XOnW3t)zuH2vm&o)nA$-?)$dt97B&w
z9X*|+V0sz*1wXcM1$#`(&&E2h#@Ny-ymh7VfDePoJuyt4H=-8u2s5>RZp!^H<}^L*
z_PS(Izq7N`W593ay&~4Usn(QG%P)v0@jDm;06mb$K={AMb&GDw&CUJz(Y`(68em-A
zM$gI#EiEng{k2iQlUbVC_DJ0dOXVL@m<X|f;Oo1GhlhZDZv{E8kL`VZeakUhqu_>!
zH$UBj4+FJ&#hm0A2*eCd_3Hb&Isrz;5Tr3TrFAiVV$Au+aeJ;a`JP1-ji`aG_iQ9z
z0*9d@j&_~%Qw@z%tN^|+%>o`RRvTmEP05$yk#*l4V>_=-6mz!NkX7oa=5j%+gr#Fl
z5yerRXK-)pa3YAFo*r4mdm!FuK-2e`v!hVM+W4{87par@*IMnZoGv}XmfX^=4tS5z
zt*Vz=Ok<c`=+2=9WuGw-eNijvmJ1O!UD-XGr=JOqQoP*+(#XQ!z4z%!;jtbm_F2jM
z0Dy2l0f5E|;KSQN2i{EpV1jEW5En~Eo>qHV_d+I1N=n+^-u8bvY3s%#dDB4BWXn}>
zV@vH~JP|p$roR6Ap6BWHg8cl{Sr=<oX{16>TztH2WeK4rCxuo*QE@S0l0i%k+#u=k
zG~Ud^hYxWqUmmcujxpoMTouxd?W7Mg-W+n!<%*Ox;YtRGHpouT{m6x9(RDn)zzSLj
zZS8Bw7T5a8)jjEx^#y)*_>}F~D8u74G&CrQo;v^dXmtE%V`HMWd_qf2tpY#>v^UnL
zRrZYr92;BliHU$1796o2Hsc;X)>nFIL-0sdvj<~+TYt~`tSBO3Uq!~7DABKF$nO^A
zESIhfs(+K<?fv0*!Oi?LfC|SetrMXu+#OKrXm1z!GvO5U3R==gN}l|j98>6wgI*OD
z76L$`$Bf6UFb*-hJi2}4W^ye<=~s<^!F4YDTbvPj`&02=eVm5JYdD&?Z0gvACElAY
zmrggnrSe51xty2!HW%>`Tz@{bMEopp<2tQ2^s3)Os#x9V>)vT-d19}oe#jSA%Kb-8
z454?Aj27?3OHFQ5F4=6N*<ZoDooA7S(vk{{Z{O|r2V&Zrv;GH?;y=U69sO8?vT+aJ
zX;~MNX3)OKTEa2GFVu=MS;@eojR8#UQb&)IMo1IBCKPRuOw!@-I_>tSdB7Vh&KG+1
z*S_~Rx1FP~C42sp>jhGGcl!Fv(;O>VU5DQ?NYI=UE&l<Q|AT$`KYK7LBy;g{v=cdp
zrSE9(kAlHs084E%L+rGA6TwdKQgG?sOcon9e)kw@w9>@REW0pYB@tb{z>X@TjBH<b
z<?{%UWP6`3ElDqj<?`T^TG(wtt>;<OCjYE+z81dN)x(k^=INo`-B2dtJ+EEkU!-Sl
z42=dwatNIl$AzV5ZakEcou>J@xj0E{A5o_tD0Y^{NU-9we50ollh=Q{-J*M-V5RD_
zAsJ*eSd4?nG<$iJc{|A6bNZ)s#(HgWNm5L{D;q&unsD~$-YreNoXJZ4bsYeD%F`|+
z@elfIEYrPJ0jj=zRQ4+Ut*sP7>@Ma0Qa0hN^TD*4y!>E3HbMsryAh8*TlQ;rel1Im
zksCE4g($#)0Drb&K3bxFKZ~LL?uy7(7=7hsOhgLZ(hyzS7f(S$hALSyuCW6B%VN$<
zTZc^*>Ngx~tJMWpGAK`ImoR7rPWILgUDWrTM5Z@mEw(4AD^V{Qp_;wGOvE;lRCno_
z?Ac#IDd1zkyG33oWL@;D4%8^nVCa7sJiBy1%H*cc8FMQqRfnv1(a;Qk6!P<kt)8}W
zZS9yt-~4auyN1~-{Et{sT=*o9(<5g2Wn(aP9`@Dc1}`4^CmwCDW{Hgw8Dgt;T>i$1
zk1E4hZ(!;o*(vP0ZH~H*i93drXcFB1lJA$xzB4o9F(t~<)cF*x*I#R>S4<ugdr``H
z?qb3_??lbuKU%(qw*-|uc47xOpNOAD9x9HU23z330(hCUyo|A~TN?5T3`E89*y|6{
z$v!xv&exaynqo%Pap$q?7jn8wKASe$38YeYGt*2cJd73BGjBVZ^Q^TxCA&S6P2}ip
zerH%{7DLNgt9Ish=h-b?$uA^%@09758VyQlaUgJ5!bD0&biHUb(ia(Iw`tN`+{}tF
zR-?xXoLurxv2wQ#-IejFg&>7{!WAQ{)4=p#4NH+m&|2a;)KA7R9aW>y0YR+2vs*RR
zXg6Y!@HXN;;w{aPJ@<ph<<Q(2_th_uN!4DIUC&&k?_AwGvP%5WZ&{)9Y_kpmd(TrH
z{rIVm2>W2j6Ru<-Y9%oZ_r|H8Bck5<e&alRU1%fRzMjX{HYkN6Y1qw(s{?fbO@WHX
z+R|2G(gSUo<F8`#-kqX(5Qy)1mw4*ijL%9&r_@VUvqtjtFC+Ykbw;R#5n?PsnbCwD
z6TcnBKS~|R*dQzkkh{u1D=)uO{bnfidw*_@hGqX~ko^)O3qPRg9SQx@;Wl|5w?xe!
zYOEI#SwwKm<YAFyU`0j7=Sv8p&_T;(t&oP-C#(K@r#<iMXN?Do;8Mi`!3Hi$;kYN2
zA19hQ;@Q4E9QyggFB3=8S*{#y(Z2rc%8gBzm3-IxgA#o*q3-GT%2-vXEkC<Awzp+;
zRTE4X<|<m<V*F)g-GiymhsY#s*D!F>y%|*RF5YIm6ar|{<>Bx_$~Zoqgj`)%TOR@N
z#w)M*n>;77B~nfAFb>~661mH(xE_`pEaHiH9GR3fw>lUPS6hx4tm{~Y2>(k@78+vw
zj5LIUoJ0bjCE;TgUL^{{KSZKWQgXS<W)4pYDN&<(yF!;YSDz5UeKmlZvU@$hwlmk~
ztun$AkP>$shv3MD>m^ba41Z~HxXPpC_wR;NYKa%tUzj$AgwGd}5>Na}ByC0S&7cFW
z@*#F?1SQ5f&s~eU;G?art*#zVFB_(g#l7_+hDnxCS*;%FKg{mjcR7I5b9ZTQaF9Vh
zVZy0b7<f<B8rB!C!PIsuZ!znc_5^imMt?u9r}F<Q=(NyXs9kQ}o35{;!(0uU=A88f
zS{5qp?VuXHV!ml_RbNZsfUSoMiKH9=Q-xlk0j`aj?#t{Jt6t3}kXmQ+iH38X^s+}l
z0Goe%Qw!7{P6DOCQuFD)ZkpnsZ-N_03AklB(9r4y8dCOoVvT>~P*%2n_IYbus+bRM
zrS`Z2d|KH%Dl02nEoq)wca@LfhI+ZTI^smd2hUQ{bi{1hr<5k(@;yiPVxV}s@A_CJ
zD~gbkn@LnO1IYy(95Bi)#-$1=+{=MiL|&*>zP-f5!o)<uZ=Xqv6NC(J<8qw)Dj7;R
z={o-GwGQ*+-sKUKTUk3-a-Dup)UK?otQOZ71HX5A{D)lXyza9nu4YGz+zi@X8{M(<
zJ01rrIv@jhN^(}MS4l}VMy(f9d-hNNOp{0e{~<3Yhb{i2%nSt*fEdm-&L^MNM$2iu
z)=FQfUH6P`0xs0h_we2LI%CjPGBV%o&fCJCKkWr9gG#rWuiW>tj?iwCL!?Lo*Q{&T
zxVKi`Yi?%db9T5$)p4#7SdSQ%ge1Lw?eKfzK6<%r+SMmQ8kmWjbaPAn*_!$;@e}-3
zgSh~U$**43VkN6(U}0H3T+E1uCHS*+yt{l4h{HRo@6|LkwuMAJw&zl9>+Yz82MuRd
z_kU#JE0`{gbsB$%&wFd>IdAjHcKE2x+1c^N4Z@DTx021rzw6`T;^t9_sAy^-_m8lC
zR2?3T2DR^UbD5qH0wD{8ZGv4dut&FU-C|=a%FkZ`*4voE95f5qFfY|AU7}>rOR$nh
zOW9^-W*P<YE$`&sZgEuruQ<2t4AKeGgDiZ$#P4(!z?_saD>L)TSf#a<RsIAzJvH@s
z%4?$$z64n~!~Oe*u-wtn(S|klGqCA?i-yukhA5;7?7_p@jaRM1Hm-U$4y<`hY^)Rd
z^v|F5@jsJp)2AcNXQX*nt3nNqK%0VMQaQffD{{OB%F4C8d7zRq_3K?%M@q=C#ba_@
zR)>LU-4tpDa9L2&GzME-zwVrYW(<N=6h~qMv+aGj4S)C}>oNLl8Xdcs*>CB4FcS(i
zc!C*_+$ra*{o($&on7rbS!z#}{mkd<b`5t`hC-&YMPW-ri-A6#MxWEk5GEzfXP%CZ
zju|38H!IZ;{+1(eB|`zDcQv<Ja*J$!!a8GvHC7VZ3i6YG)JG!0N`?Y4>68ZQ!JjEF
zyD#ztH+IkA{X((@+-3jw<Z&Chp{a$1-Dueb$P>s&3Af2FU%p%=Coj<|&4OoVMUjz_
zwYSaGH6jqocXp{yVS$8=zq}YNx9~gN8`%K)BVWA$80s8*GXw%j^fqwHl&eU$TCr*J
zoqXa;lH0#NT}Svck=CuY+g49QYJn{D^y$+Zg02cmN@$?k^ZxC+KT9B0*;MKPCk_Q^
zjtt1qC+Y|Un@*3IdF0Iul7fal@r|8B+uHerhzR`P0_facT)wf(H|abec}3WilVZ-~
z0GAFJe)!;#T|^*W6A3Kw*-k(No?cn=g9V3pH+q+k3D@D?*mW!vq>8#s&YtuBy;Y@7
zFbWPo%yTMC1&aM$P7V~_nrGuxAVY-3@L(V+m{~LmAzJ0dOyKJq85uoTpb=OaDb(IU
zaWuby78BwMc+rW9r?El;A_&A_FiUNw;h!SgRyT!rMaoJ-Ui&tiMna(IoXo{4-Bi(9
zDozc&bY;r-cpdt1B=Pm@f{?Nrz5W<K^hEgAajL7gD_NkbYhx})38EGFe{U;oxeKYj
zSGezURMBZFCojL_+2{wG0(x1<K*A8B=l|CcO+$0van2&cM`Peeg_EY_B_Gfl4ZYSY
z@0zqjoF}68K)VoyxO$W%h(NR{IGsu57;@ZbJ8_L9lE860%JKMCPke0M7e0)=s883P
z;|=OYY)p)S$2{qwM3LyprlhoVyLCyUD?|yT4uvF6qvQQ`a(2B<yEV`lNLKLSeN4H}
zg6zk|#Why()amTx5LCrq0U#Y<==o7<JEgL*43r%G`V$&5Nq6c>5&k7uaqRrPOf`52
zr2SU%J0hXW0qyBJpKSvuY7mf%&I0pKC4eYT(@gRNp@#+<Bv>c|mmzgX==I&K3%wc5
zerVs=&Rs}@DSq{qdS9gxh!mV>G1il=8mrBQk~j`iB4ddLXxtB_rKRv~A1W$Br>Ll;
zlmXk8XGHJbvoUV8__`$gk)=#>Y$JFU$V9J8tj@C}<e?X4k|&P3YHDg)TKu5tp-(sb
zK+kJ}{J&y_jmS)`K6aV<J;LLC1iM{i|A&d+K`RZf-P8n}6cG_&*L27zdj3Io(D*g{
zu3K$_>~!6>rlq+Q<f}(atj>KL+GX@~bZ;vv;*63&I0l8`T+*W35fBwMG_Z$Cy4aoU
zlnhU`=>{Q(KA4R@V^b12q^FyM_nEk1J6=_^@HVBL5$DG4;<vQxJXUchJH>6rwd+;m
zifIC*j$Kf#K;M*j(QjF}EQ&YE504kmqDd@qwv&N=;C=N*hDHHj>L>`Vt|3Om0hBK9
z=B2am*614rbkaYqY=6vyDoGc;Y;@?Dy23+}(JT4T+gk`%AT-K|Lch=D)3pUHJw3eu
zgXsyEbLN5K7s1|)5ruRNePGMa>0A=*+KX%j+m=%X=_A^#1QW=LMfQ{)bDdX};|K)J
zt#dEHkf>Tf|3l}xSJx#r_oBK*lvnMg8qCF!Hw9Z07@hm3Jy(ljZ!7Uq;4d#P1C%(|
z;2_7lot=1;k5cu`7?8{>ei$$=)qB77{1Wvm^zFXYJyKHA*ZKr$BP{$>O=IQr{NHSN
z$8DZps_+z|zKwd&5YqmOig0SR@=`nz3CXNIzqd$+pNWf0W$&VDqxr9$Yq+JOHZjt(
z(}g3e+6gZFR2Z@HF=`O`GmcVKc;^JlInK^29F$ktw273?W(q4FvsPmZb0o;W)c9@D
zNnP`DC2GNA%90yIm5T~%d!9E4Lsw{+BT;RhLccT5c^ptN#IOaa2unt;Z>4A_M8aoH
zNH?xeyxDfL4pIZS2I}6=|E}L)3{rT*@9^u(%QMx5`FWqC<qxeP1mn$qe(<I_1}MnM
zZ~KWRCML?99IM9AD__MjXF<ylH?Yda(95cv3ce|-kND|AB=x$&|CDf{nk#2a^oQO8
z1A#%}0&+|0qf5h2x)HU5XU5_x=p&&P4t`V_-E`V3-N<czWz&^|O`HH^16@TB0IeXx
zqn~0W^8Z8e<^ne)P@R>3luTS)+yL4J*ycXR`;D_vf-5Q8EthC^n+5Hr8h*kyek{LK
zYD;)SfExf7?DyLl-Z6`ZBZhv$KR*~D5HH98g}|mi1}%1OZf<R|{!chZ)7Nj`rXT@B
z-Mh+8h(LTM_ly={WPA^ffVMDF@JN_ySAG1*dBA@gZh!|j*nlu2x+UG6E_@95(04J-
zO`}K$G)cx_77&+SK&UXrJ_Q;NAO>aRd#A;3B)1i9#(pH7*D%P_5iehMqEI__6TtCk
z6f~~n7qFuz`t<n%Xd|eF1$cN|I%61SJEHyehBR+!up(TdYK4Pd#rkFsO03Ss<k&G6
z@T#eYtISz6!JS(}uUz8(4HyN`#GVKr)KG9R6bvj+c`V+L`1<C))47KLV6IV)vHz0-
z=Ji>{42GZy``x+E5rZvc5kq5g_obGb1}ywOYLZBVbm%PPd>9-jC<7fiy2A<dbnD{U
z_SL<!lbzo45^rq$YoRCA4WJ*m8jAJni6ubL)CW3KS^0HBf;7${DBo4Klb$HHYCVtn
zc!EN<U1}e3@fNUV%+D~hk)Z%@T)81I*f(^+@{tG!+9x71c>Z}SK4LQcPSfVDo!^x=
zhdlGRbcW;k)a(M9JAJwgi4r34rq{vK;RsSR=<xMWJE55XmxLXQX0bOwE08I-zmO=c
zO*i{3e}8wGj1>fP?FRR|bf(a1z=hlKng)k(0@Z!ToaZipd^X0bH{ivIn0`g3DQaqI
z9dzE#P)Ox1+Ty@OoVqt>lyDy~6vTJ-S<-J?^F?lszkBcN^&<N4S<Wl$rf;gM=gKy<
zu&1f1snGV<Ta=Lt)9AB50T(V$K{Dy??gq{g65tt7cje|iR?|(5-efEATktMGy@Bsq
z<~ZM#((;9!WB1p8kB3e5ILT;pUcTI#R?{H<jFd>PlUSdT64zGJBQkW-eMMi-<nyfH
zX3z=G-Ls8n#g<>GBBZmuCA0gH67;g$?m|4(l{(LQK8b7-Wc1z^9fK+Xp|CbqSwJU`
zJLxjJb9lC$*sKRh4qQFp?b+}{dr2-_=)-g2Te&hnkCaf<8+{utVxMVlDa;>|oSI%%
zDS0e>qx?hFv&UVb&K)=06q32jfW|(rlsDWa9AN7!!jZj|A;*!PG`A@*CD;}W3HY6z
zydn?8JBM^&F2JP&glpsN?fn%Oks0#Z43|!TDgZxdZUvKL$o?7>+=NTMG=i>$8U+kN
zLI5-WO#At%sC=Jl^b$Sz83<+;_#7M@sD4mvgt`qXH3vtp4eth^0_gw1^h745eBIgM
z#gn&zz6YB#aIo{+x8hnI8?Z6Jr&SZoV(cYu8&lhnfET*><}SCwq5$}P%M;s~NLbPj
zB_8?}R5eJ>qpvOm6CSbYR#{e#`9THaVY?!7KzlI<8}UN=Z6I|iAwn!NWPcv~0daqQ
zUd7*94=@m!?4TlBCHCR`$i9>tIEas>ijfPQG4e&d{kPrz`iSvoT1~GU)C%+o#cF`G
zdS0RH_sVYbcC@vdje0eO|0a-}m>~38*_uRCGl9f}GgoUQ*7IySUVn%Mo0iLdkU)&q
zZNoRKzfiZi{qB+PT?X}aN5tn~VNZWr{9C5xMCmeJSU3Jjbyi5*%GS-1gf)m%UdCX^
zjHb>UD#iU=bFZyF?c1cHMJ;VZ%9|WlslJAfGFSRmyj7o0+qiwLwlHKaA^w6=y3!Bc
zwZ0W>n|)UHNGU;VVX=ap4lR^exNMW~a+tenAcruXGM>*aqA2R68F`ib4TtS+WccAr
z7JGfmqnT8yOrGzY3RE2oTW`jK{fYnio&KF8?q3c9C+vTOH<7+RM2IGf9ba`8kSZPC
z#P^~qE;raXDGT$o_#xdoZG2azrYkoX*pj`pZFfzLu~S|5P}JgOD!l;~*}GKaoMsO-
zS1NFhD1jxE+YBZAXqLz=il(Qpe666I;1J0AUc5Q7i#GhPtQSC=t4{o=Pv(jk*6%ro
zRz4Pr>TpqmOv8G)b;nFI@jk3Re?hOh1zx`f{}&V<6d+e!%6uL-ShY~aaWD0WnA~Qd
z5*C7>p;ogFZJ(VpaQJ<!){ZjE^?qL&aF<Lo%PxTw8B6PvN)bwr>&Yy&|30hg%%)t`
zHP0*8Le3>xD}rJ+4yH`y^dzICpRU3B?;}D=!oSnYvP%L7m_7XN{sV6!1rre=s@AV0
zPHaE3yy--R{;hJZFh?^TJhzK_Y^AJ*<9W{Qg{=u`CooWTfV3Nl-)d>a7c*TI1a~bJ
z>gDXd7`n4OU}0TZUPv#C!SPxqCSR)bc*bP!8?R;wI!p2xXGAB^2!{@QCHbc-4pijZ
z?-Fin{CpR6aC5F#<?9ut%@nH#%IR9!u6?MNoxE$4uWxJkNj}j|Fq2H^*lFM*>gHVW
zljLI|4VaqUdT>II6G(f|U{HdZpZ>Z7Ki!6G>4R_re(oFwjFZ0K^7K${2wPlu&v0Db
z@#PCYRn$)>%<B((IeV(DfOf=jsMS+hcY8dgA;~_n?plA-CpTl1%ANP|7tLgs`pCED
zsGSC;Fpy91z~H02p2;T{GqSoj#IAH!OkhB+d=+TSmFoe+i?i5tHTc>6ORn|t_l?44
z`C~iP(9X)2E!y*0qh^2MSDuMhraW@J39Lb-944ac@zgyqZK|#J`{`=Cbg$>tf;K*-
zF>lY0$%t~RMfr%hVI7eiB}I%-5)c&=BgC}TE}Tkb*moU!jqn1A|I5`2OR&Cv$%VD`
z4W8ANVcAa}a|q;KF5c1c?Ci<Lwh#>oiK>RidkxSh*f9Q6v%vqPp<?8_Oo{25x&l@)
zL;HbuL~F|!vlWlyo4K7b-c-MAQ%*UZXF&zn2+I~7&%V`>vqTmVU9@vnCfogz$*Mb<
zqaB>>W16F`1XOFQjxN|NaN-P{C@6DQ+u7UZ4lYw{h-}2*U#n^HJGV(ORNRUD^OF5&
z8kc2<xd)Pr<8iSb?+nNdlo3cu5PWZna*U&k`z=kihi$7t&LAU0>tS+}1Dz-ZhN@{E
zM=DgA_qsH~lJ4X63;*6!MdLG9N@_L{ytAj+o+gO;QNS%jc%ed`kz~=4sQEd~gKXJ{
zy)W$J5j_zOwG`C_7_s+Xc93xT;?h01Nvx{g7WQj{jWW`dTdJmO7(<-K!)95k=S+A{
z={8=_Y`}-B9_b|5!y7-P=<?VCsI#ILKJ|{R>#k%d62GLswn%=F`<O!7@EB=8cYEZ%
zv-=#LjTCbC^&)C{3AOxV?t^ZT#)rw@d!Dzni2b*PM)tS##y@l_0oen#012#i?jX;%
z(f|7$F?!hSLS+Eyc&vwkG&!d;SdUlvIsj%37wMAI(D;E}@LFX1W+IrNACdf~ZtOAF
zEi>)Rk&X2hnFe$*-~RnOILi_hlQ`+_-E#uttPweQ%pQ^3_5t8IFMwp>X5bU_+CT9;
z8~|<;cq5==oL7g7fInNi33Y7w^Lib)p~`iNf?n0s@Tc}L@-6)R)aeyRi=NfMJsf9&
z>!|VEvy%7<F4fe81hNIk>j6d-31;)#SvaUN@IdDKXwM~lUfdxfCMTDv&_SDr7*T|P
z`NV37bjwXU=D~BL01JpSdRc+AZ_UcLa2yGA5?<O|i=3b3FCXn5WY_g|3mB17)idTw
z)>*swxy;7KrjYA;08<zqtCZ9u{3F1FMFx!m!orhQm0g+}q}bvmBO?m-&|14qd02dX
zMF(9>tgdtf1ypvwb59!d0eJP>M-EgUz!>+{xRgIlF{AgxoW!+6dA^N6@EPFrTQ%OB
zrme(^G@<~11MjluPa<Q>gbWLCm}2VOJY&B%G$bI4mtNHJ<Q0AAyyGFCMtGK9;EH$&
ziB$B!ok8~0MUTK1WCJ@`splOle-MDr8zg$2<nX|9)`DNQVVEftyMQ@BhVEN-2AbeU
z#Zw?7ps8SCk@p=mF9$^Fbxw6#Sy}a{jsbB9&Sr^+;82UX^WNTAI*1lWF0qI6SEkV;
z-ztL=3ySt-=OaEmR+%`R@E6T`g|tC$is)stm@ZZvL&8o{7hI~0xq9-qQ_gv{n)Ug1
z(Oqamxx{-N-e7{aha?c%$F7d=An~*6R9FJ+k%^*87V_YtGd<c}9x?DL8nv1FXoN+V
z^C*=!@FGFjleRuqe#NidR@>za*{_YXTWq7BNFk7SX7>BbSp}bsRY1EEeeGVmDHoAM
z2oUl~N$zc=nqNu2Mj{ZI2{=?Ty$*V|wzfT!4GlsP55Y`$;B*n;zd6&b_3!K#df1R6
zE$0(*D%C3q8NPcpWHCWgjh0Sif<{AKV~c>l!STgt;nB=~-xyTemAR{Jhi1HJvZ!)Y
z==WnjIpSqoPS(O)n}@#Bro2ya1E{Y8X2?~g>R{_F<k1TqwIP%0pRtGfFGd2e=>NZ^
zjiQm*pSSm_;g414p$RjJ5HoFYBqpQa#@>w`_k}q1LBZDBB@YiwBJO`y4+sDxP*K}U
zxtf6z>RsLQte^Bv{-9@YDUF@j<+;`0Uyfz|X$F_E0;B4M*Ms?&)Ive>`M>SB+v?(o
zF(W;4eNFzw;SE=KIBkMic7e@#Kr5Fnyx>U5n{CZLN*QK3Xl*UCxm;4q9yU~{#gr!%
zoVn;>5JP`~jNY#aWNJ)Q7;T7=d%6XMG&n+G^1;P$1+MP5l=xN^?3X(yW?`N9;i!Iq
zdpjgk0b6EdVHlekyRl=I2wVL5g)JeFDD~MB&u;>}&z3uY3-EnnqQJ_=yswhlv*4cq
zH~Vie0bFfnarAIefquGlHe?>Qs)agmbyklTDQMVJS#Tv^)q17Fgbsz;|1V(lzs+p?
ze>zBdkq7HBk*;Td%|YyW4%s^F-e@E>CJY*+RE;!SCrOWb?1JdQ6u(jwZW-woaVj61
zVarXayS)^Tprc!Q8$ZK;m$!IXi7)jfQy8Jl93G~n>;Dn^*8kYDR^keRgJ}CYV~iGn
zgoi@5?em3udMz?qjG4;cEpa=fsvcUF!dfVQs<EWj)loq^6W=Y=qn=<z8W&IgqV!GV
z`_uLH>LoVn0BTx1wP>%3&fb3-e{Tlti8+h^Px>-$B6oov@t%*@BxBcO3af1Yg_PQY
zt`gVp3|>@}ec+UNIh%HwDX4BD8FuFr16PVbE&;a6=o=02Hdp_<O1GBx6ud{kV?R<<
zBy>8z%X@oJ;{LV7X$&;kjFyetHT!_{#={1L@yyCl0WKYoJhoutI`=<^7-=QMqrz{p
zqI6ivWMc-|bBHJ)1Ym3Bs{0#naG}a#=$D-Z%*e=zw(ChEuS@(bSQ28@_k1Al1A76k
zFBebGx*;o4?B~dvH~|JFi8PJwTVVbt-`U<i$>0Lf16oiL35b1lb#-?$KAaARFgBI|
zR<iJc<=sPXEAsC+|K_UGM(4anl9QA7W&NnP;iRCKb*$CvVr9SSp3X;suUt(BC0$KZ
z6SehyLMuVtflmqpDkk*1kn5M6>p+yS8LN1SRMFB(ijQXyt+SmJsGmHS*g(fJf-eM4
zqbqbd=k;g23Wj+S6Iqgt`{R?i#F1GAHp*AE6V%V0^K7y)Okw?dzd7-5$*MRHvF;pt
zcVP?59P?<w^Zt!EVhLhmVz8=nLsNZACQ2b)C>hD+bF>S5oXS+KAdrq={l2NijzDm0
z)cEYOq7+kk*FjB!?)`4ELeMLi3W6kAdwmi7Vc^z$DVB^t^r;cme7@S@P+MEe<91rV
zU@ZXiUt(+#+cIuHY<SIlXCb{WJ9nWi9uR#P$QxMJ;!RQ1+4pRW0isz4UXC|H7Ah)N
z_`bW)3bo-Nb{13Jq-|hJrYl{fN%c5ztpGB-gQ77nt3*C1fte6kHMMnS%Jt&<-(%$#
z(PQe|-+(rmcAq7vEO`7`^lU2{Fp(CVVF6MIoUc(ICVidD^nBK7usDFYzhWE*;1}|_
z$PpO}BcnM06D!SrOOUXgjE%G4Zi)s527@==PjUlQSgyqizvPD}1<@%MemMq$rvsBr
zPA;^n2!x!(-#M@vm^zYOr%VBG<#V#t0i1Oz@3R#+(<Fl(=ks8mYBW&0VA?Mj0^#pF
zmDEMudRI17*>zr<hw}2)pWW{OA#?^(1|bzc48{z@1QEAPa0|?YfLNyse6oQ-dR*LK
zMe;eT7EaD_)=G1R938vK(#^wm=ML~Za>~m6FzNvB^iGEIkBQF{NImuu0IF^PA;1&I
z&Bv!%x@8ZCO@aDizH?_8I2u#yY?X@N{XniB{cH&YAv?dI7`E&`#>ZG$SYYqmR(xY$
zKN*BnzN(Ziv~15Gka14rR8&-e!EFw>IlW4&cw;jfe*2#wlP>pXhZ>ZKx~#;Z=3yWw
zjur<j`j7%*Kjh`+@;fakeEncIQ3D!>A#mHqMn-RqeQbf*7l9El6%`c#SWlnk^_YXx
zAYx#-u(((R2H5mM;8qU59_U*X)KC6C*I>HTu19f>I0KLbNVy0a;JN-Ed7Y10)hX%v
zXJDovsE>aJUYT+k(HA}4jpgEr{m6$c@fB2>?VX+Ye7+T_RqE5N*$&`JljGtlz-s_>
zv!<a>yipPa{l;|j^j|Tz0;ne-DPiF1;Api(RaF(scZ!<=A9gs1B%-3tIXMqX8)u;-
z*c3v;<d^I6cQa74AfK_Zv7LgD{g0U_AXdTn31XF@ipn;~(uSa(Vkm7)HP%aK-}i>4
zYz+ma$lT0o3ScEjm=Fc<z~EB&9DLa$9SzR`BY%n+BB!7@Ry-TukK~)O7f3@Tu+_tu
zUZ7haI7z2qe**zmxhZ_aIl_JS?%jL$RI~eSCTi~9xpPNQQ1^LC>jV-)9)^sb@s7h}
z(@3TD$l##b1uI4)JDi|bS42<jKlx*1F&RfU(Pg<rpTRDWd-xET+SeQ4FNlPFt65?I
z$o>Y9w5>Bg8ok^=50D1KBMfi=z1obSRISAd<iDfR*9`QDMO)XmZ{L0o6=?QMf}}MC
ze-OVl0)$@(GDpqtUtW@&&t`>Zfiv^!<}Qr;eg6CzNLfn!i)I(SZU*W3><z&hZ*6UD
zPeR^}rnq}kM8vl{3AD_JJ4{Skl~zNrXaI^w>)kd$aK0$i*wAnSRHlBNv-0~f`=HwQ
zkRQz`7Fw@}f=7T7Y#m%Yp|x}rFsamX?m?jz^?kD#FL;roo4^S6CQQ4Smur^>WZz&2
z)f}b@VTdk;#OdL22s?zFP3Hm5GQ{qts(DXJ+s;X<T~pq#Yk`iVju`l@FS4|>wGYA3
z@;9MkN)N`;KZU&qnH2E)HW)Nq)_&Yo-#42}xfb#oMJ836mGuj>Kq2={+P`8a#sCK(
zd?50{a1Tze8GEJyryE3(`U4=_ksEqDz&nW++Qw<+>>^j7Y^I|U_T*+~w}n>fT>A$3
zpw4Zh*uJH#-4+K_>Ll)`-~SHcedw4~UW;2>A}#WQ)a}_gg;JpPn>lA|p&ZA1F`wQL
za(`u%1Ybn2V2IhH=eZ~9{B=z?qkaBNYHDbZ*(>t4a>53KY0t7zo5kJ?JBWgN<i{)t
zX2R6HW{h7^C{%Yhm8dU3y(_<~Wfc{t>ReXCNLgT`#;skj4S?XB?}`TpNaNnHu9ULc
z8hIQiUl6ow!1sS&-J|BWhpD*6`z%EaZwSP%w(aijf^7tR8P6b9gG<DA!T2c>zPwPc
zg_+s<X!&g(o}nk4UqgH785lZmmqY(;VQM;I-%5Z!=n$Pl>tdxFsH-#3+!DumI^A??
zc(zS$*3|^Tk;q}l{%6kI)buNw6_r*C#t=w)Fcx|WNtT&~#jbw6>Sx+?^VwcWI2jx9
zU-`F%eA<$6bab>pqlk+FA9UscOYUu!i6$Q}4`d7sDPFn8Wjb1B1|FjoNG>C_PK*8@
zrX4zN1R-J66*fYbX8P08&w8rLS4T<=ISlU~zE8DPS)h6ckMDbWlqTwD;JXVx74^&~
z+WglsI~<@!@zc#g>g7!>dOwcGo?ZEdvl0v?Ko+FbgBoai7?>8HkYGF2P`hC3>FL><
zF05Yo;kibkma^0M6V3;-ot*?%Bdx;--nv+^0YvVzlCnO>b==5*n+keCc;Ui@48OD5
zF?GZP2JqZl!u#y)?xsICY@H9;0`099Hr&ecHa<zv%~ndPO@_vY`^hK&pdf6IiBhn6
zL)cG(5emPxP`|zkgzN3(TtGpnL=LxFz3K_a%f*`bkFmUV`?B;wR1!!lhD78z@(6S#
z2)_pKt-!Dsh~7pkgD}3C2LqGmZVDHdJ(#KlJA;qNG>k!mfF)X>FDNGeJTsJR9da#z
zCB6xJ$w%_Gw!b0xXnb~IN*RWsWgz8P_MN+MdQ$l~V35_S5~h7Y23vp$UG)+x|AY(r
zd`L5^Z-a~ciq!rQMid@uQz_vm1UCdKa0AP+`tZDVk!CBD9IR|?J*HfdG@`yI2U{`p
z5Px^{D8#~nw72XBkrsA7rK3X<n7j6t`e)(24zKP3`9nZ0O_4biHViWtEN@}r)OqPK
z)LckGki*#Z>j1?K;xoB|QjeN7nun4yKOcGR-usUqUzv?wyJrr`TVv=mh7K~~?+D{=
j|Gyta{|{|VE&7VN39-qo`i2saND#78iVun<o&@|C6iQca