interpolate_to_heights drops nodes in largest size bracket

[aornugent]

I discovered some unexpected behaviour in interpolate_to_heights. When using a coarse grid, individuals in the the largest class are removed, where instead I expected to keep them.

It's not quite right to represent their density at the highest interpolation point, so I think we need to append the largest individual in each time step.

Here is a pared down reprex using pieces of the code from interpolate_to_heights

# helper from `interpolate_to_heights`
f <- function(x, y, xout) {
  y_pred <- stats::spline(x, y, xout=xout, method="natural")$y
  y_pred[!dplyr::between(xout, min(x), max(x))] <- NA
  y_pred
}

# interpolation grid - expect to include density between 3-4
pts <- 1:4

# but nodes in large size bracket are not represented by final solution
data <- data.frame(step = 1,
                   species = 1,
                   time = 10,
                   height = c(0.1, 1.2, 2.3, 3.3),
                   density = c(0.1, 0.1, 0.1, 0.2))

data %>%
  dplyr::group_by(species, time, step) %>%
  dplyr::summarise(
    dplyr::across(where(is.double), ~f(height, .x, xout=pts)),
    .groups = "keep") 

# A tibble: 4 × 5
# Groups:   species, time, step [1]
  species  time  step height density
    <dbl> <dbl> <dbl>  <dbl>   <dbl>
1       1    10     1      1  0.102 
2       1    10     1      2  0.0914
3       1    10     1      3  0.163 
4       1    10     1     NA NA