Error from the `var =...` in `define_population` and `define_observation`

Reproduce the supplied example, focusing on the var arguments passed to define_population() and define_observation(). Read those function entry points and the related metadata-building path to identify why including var causes the failure; done means the example works without removing either var argument.

The following code doesn't work. If I get rid of var = ... from define_population() and define_observation(), it works.

adsl <- forestly_adsl # SITEID is included in ADSL
adae <- forestly_adae # SITEID is not included in ADAE

adsl$TRTA <- factor(forestly_adsl$TRT01A,
                    levels = c("Xanomeline Low Dose", "Placebo"),
                    labels = c("Low Dose", "Placebo")
)
adae$TRTA <- factor(forestly_adae$TRTA,
                    levels = c("Xanomeline Low Dose", "Placebo"),
                    labels = c("Low Dose", "Placebo")
)

meta <- meta_adam(population = adsl, observation = adae) |>
  define_plan(plan = plan(
    analysis = "ae_forestly",
    population = "apat",
    observation = "apat",
    parameter = "any;drug-related"
  )) |>
  define_analysis(name = "ae_forestly", label = "Interactive Forest Plot") |>
  define_population(
    name = "apat", group = "TRTA", id = "USUBJID",
    subset = SAFFL == "Y", label = "All Patient as Treated",
    var = c("USUBJID", "SAFFL", "SITEID", "AGE", "SEX")
  ) |>
  define_observation(
    name = "apat", group = "TRTA",
    subset = SAFFL == "Y", label = "All Patient as Treated",
    var = c("USUBJID", "SAFFL","TRTA", "AEDECOD", "AEBODSYS", "AREL")
  ) |>
  define_parameter(
    name = "any",
    subset = NULL,
    label = "Any AEs",
    var = "AEDECOD", soc = "AEBODSYS"
  ) |>
  define_parameter(
    name = "drug-related",
    subset = toupper(AREL) == "RELATED",
    label = "Drug-related AEs",
    var = "AEDECOD", soc = "AEBODSYS"
  ) |>
  meta_build()

meta |>
  prepare_ae_forestly(ae_listing_display = c("USUBJID", "SITEID", "SEX", "RACE", "AGE")) |>
  format_ae_forestly() |>
  ae_forestly()