validate by one analyst: confirm there are precisely the columns used across all analyses

Debug why this validation doesn't work -- probably time to #55

This code will be less complex without carrying the column index.

# scratchpad validation

# validate_analysts

# source_dat <- targets::tar_read(dat_eucalyptus)
# source_column_index <- targets::tar_read(column_index_eucalyptus)
# analysts <- targets::tar_read(analysts_eucalyptus)
# set.seed(45)

get_validate_analysts <- function(dat, source_column_index, analysts) {

message("select an analyst to validate")
analyst_sample <- 
  analysts |>
    dplyr::filter(n_analyses > 1) |>
    dplyr::sample_n(1)
message("analyst ID: ")
message(analyst_sample$analyst_id)
message("number of analyses conducted: ")
message(analyst_sample$n_analyses)

source_dat_rows <-
  dat |>
    dplyr::filter(analyst_id == analyst_sample$analyst_id) 

message("check the number of rows in the source data 
is the same as the number of analyses recorded")
message(nrow(source_dat_rows) == analyst_sample$n_analyses)

# select the columns that are not all NA
source_dat_cols <- source_dat_rows |>
  dplyr::select(
    where(~!all(is.na(.)))
  ) |>
  colnames()


# match the column indexes to the source columns
# and extract

identified_cols <- tibble::tibble(
  col_index = analyst_sample |> dplyr::pull(source_cols) |> unlist()
) |>
  dplyr::left_join(
    source_column_index,
    by = "col_index"
  ) |>
  dplyr::pull(col_name)

message("check the source columns identified match those in the source data")
# exclude first 6 metadata columns
message(all(sort(identified_cols) == sort(source_dat_cols[-c(1:6)])))

}