format_ae_specific fails in single group tables

The function fails when producing a table with only a single group. For reproducibility see example code at the end.

This only fails when a total column is not included. I believe this is due to a larger edge case issue: when statistics are calculated they are stored as a matrix in the meta data, but if there is only a single group in the analysis R will default to storing them as a vector instead. The format_ae_specific function (and possibly others) then fail because they make use of apply, which requires a matrix. When 'total' is included in display it gives two identical groups, so a matrix is produced.

I am new to metalite, is there a preferred way of generating single group tables?

Example

adsl <- forestly::forestly_adsl
adae <- forestly::forestly_adae

# For the example, modify the treatment columns to only have a single group
adsl$TRT01A <- factor(
  adsl$TRT01A,
  levels = c("Xanomeline Low Dose", "Placebo"),
  labels = c("Blinded", "Blinded")
)
adae$TRTA <- factor(
  adae$TRTA,
  levels = c("Xanomeline Low Dose", "Placebo"),
  labels = c("Blinded", "Blinded")
)

analysis_plan <- metalite::plan(
  analysis = "ae_specific",
  population = "apat",
  observation = "wk12",
  parameter = "rel"
)

analysis_plan <- metalite::plan(
  analysis = "ae_specific",
  population = "apat",
  observation = "wk12",
  parameter = "rel"
)
meta <- metalite::meta_adam(observation = adae, population = adsl) |>
  metalite::define_plan(analysis_plan) |>
  metalite::define_population(
    name = "apat",
    var = c("USUBJID", "SAFFL", "TRT01A", "SITEID", "SEX", "RACE", "AGE"),
    group = "TRT01A",
    subset = SAFFL == "Y",
    label = "All Participants as Treated"
  ) |>
  metalite::define_observation(
    name = "wk12",
    var = c(
      "USUBJID", "SAFFL", "TRTA", "SEX", "AEDECOD", "AEBODSYS",
      "AEREL", "AESER", "AEOUT", "AEACN", "AESDTH", "ASTDT", "AENDT"
    ),
    group = "TRTA",
    subset = SAFFL == "Y",
    label = "Weeks 0 to 12"
  ) |>
  metalite::define_parameter(
    name = "rel",
    term1 = "Drug-Related",
    term2 = "",
    subset = AEREL %in% c("POSSIBLE", "PROBABLE"),
    var = "AEDECOD",
    soc = "AEBODSYS",
    label = "Drug-related AEs"
  ) |>
  metalite::define_analysis(
    name = "ae_specific",
    title = "Participants With Drug-Related Adverse Events"
  ) |>
  metalite::meta_build()

outdata <- prepare_ae_specific(meta,
                               population = "apat",
                               observation = "wk12",
                               parameter = "rel"
)

# works with total row included
tbl <- outdata |>
  format_ae_specific(display = c('n','prop','total'))

# fails when total row is excluded.
tbl <- outdata |>
  format_ae_specific(display = c('n','prop'))

Start by running the supplied reproducible example and inspect format_ae_specific, using prepare_ae_specific as the preceding entry point. Compare the single-group output with and without the total display option, then verify that formatting succeeds when only n and prop are requested.