Issue with subprob numbering

[sbihorel]

Hi,

There is an issue in the parse_nm_files function. Subproblems are assigned an index by the following calls

  x <- dplyr::mutate(x, subprob = cumsum(duplicated(!!rlang::sym("problem")) & !is.na(!!rlang::sym("problem"))) + 1)

Let's say that a NONMEM code contains 3 $ESTIMATION records and that the 2nd does not generate any output in the cor/coi/cov/shk files, then the code above will assign the wrong number (2 instead of 3) to the last estimation/subproblem.

I believe that the code should take into account the TABLE number instead of juts summing the number of rows with TABLE N0.