Bug: `summarize_row_groups()` does not provide the correct `.ref_group`

It seems that .ref_group is not correctly populated when using summarize_row_groups() with a custom tabulation function.

In the example below, analyze() correctly provides the reference group data, while summarize_row_groups() passes a zero-row .ref_group.

Example

nrow_refgrp <- function(df, labelstr = "", .ref_group) {
  rcell(nrow(.ref_group), label = "nrow(.ref_group)")
}

lyt0 <- basic_table(show_colcounts = TRUE, top_level_section_div = " ") |>
  split_cols_by("ARM", ref_group = "B: Placebo") |>
  split_rows_by("SEX", split_fun = drop_split_levels, child_labels = "visible")

analyze()

lyt0 |>
  analyze("AGE", afun = nrow_refgrp) |>
  build_table(DM)

                     A: Drug X   B: Placebo   C: Combination
                      (N=121)     (N=106)        (N=129)    
————————————————————————————————————————————————————————————
F                                                           
  nrow(.ref_group)      56           56             56      
M                                                           
  nrow(.ref_group)      50           50             50

summarize_row_groups()

lyt0 |>
  summarize_row_groups("SEX", format = NULL, cfun = nrow_refgrp) |>
  build_table(DM)

                     A: Drug X   B: Placebo   C: Combination
                      (N=121)     (N=106)        (N=129)    
————————————————————————————————————————————————————————————
F                                                           
  nrow(.ref_group)       0           0              0       
M                                                           
  nrow(.ref_group)       0           0              0