Fix footnote issue in prepare_ae_specific()

[nanxstats]

I tentatively fixed the R CMD check issue in prepare_ae_specific() by wrapping the rtf_encode() call inside a try() call.

Although the RTF output looks ok, the root cause is not fixed. I suspect this is related to how r2rtf handles vector inputs (specifically, footnotes), perhaps relevant to Merck/r2rtf#88. Therefore, this might need a proper fix from upstream.

To reproduce the issue:

library(metalite.ae)

meta <- meta_ae_dummy()

outdata <- meta |>
  prepare_ae_specific(
    population = "apat",
    observation = "wk12",
    parameter = "rel"
  ) |>
  format_ae_specific() |> tlf_ae_specific(
    footnotes = c("Footnote 1", "Footnote 2"),
    source = "Source:  [CDISCpilot: adam-adsl; adae]",
    medra_version = "24.0",
    path_outdata = "zzz.rda",
    path_outtable = "zzz.rtf"
  )

Warning messages:
1: In matrix(text_rtf, nrow = nrow(text), ncol = ncol(text)) :
  data length differs from size of matrix: [2 != 1 x 1]
2: In matrix(text_rtf, nrow = nrow(text), ncol = ncol(text)) :
  data length differs from size of matrix: [2 != 1 x 1]
3: In matrix(text_rtf, nrow = nrow(text), ncol = ncol(text)) :
  data length differs from size of matrix: [2 != 1 x 1]
4: In matrix(text_rtf, nrow = nrow(text), ncol = ncol(text)) :
  data length differs from size of matrix: [2 != 1 x 1]

The warning can be located to the footnote handling lines. The call chain is:

tlf_ae_specific -> rtf_output ->
rtf_encode -> rtf_encode_table -> as_rtf_footnote -> rtf_table_content -> rtf_text -> matrix

To reproduce this locally, one will need to use R CMD check --as-cran which takes warnings generated by code examples and vignettes seriously.

FYI @wangben718 @LittleBeannie

[elong0527]

Wow, the issue seems more complicated than I thought. I will check from my end.

[nanxstats]

I'm closing this as it is now fixed by r2rtf v0.3.5 + ccff57b + ccbca87