Fix prediction with glmnet after fit_xy()

[hfrick]

closes #250

For coxnet models, we need to save the training data to be available at prediction time. We save the inputs to fit_xy() in a fit_xy.proportional_hazards() method. However, we saved them as-is, it worked for at fit time (since fit_xy() can deal with data frame inputs) but failed at prediction time (because survival_prob_coxnet() expects the training data as a matrix.

This PR now saves the training data as a matrix (for x) and Surv object (for y) directly in fit_xy.proportional_hazards().

This came up in tune_grid().

library(censored)
#> Loading required package: parsnip
#> Loading required package: survival

lung2 <- lung[-14, ]
lung_x <- lung2[, c("age", "ph.ecog")]
lung_y <- data.frame(surv = Surv(lung2$time, lung2$status))
lung_pred <- lung2[1:5, ]

spec <- proportional_hazards(penalty = 0.1) %>%
  set_engine("glmnet")
xy_fit <- fit_xy(spec, x = lung_x, y = lung_y)

xy_pred_time <- predict(xy_fit, new_data = lung_pred, type = "time")

^{Created on 2023-04-06 with reprex v2.0.2}

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.