add failing test for check of ascending columns

[elray1]

This PR adds a currently-failing test case for the check of ascending values. The test hub has two task groups with different quantile levels requested: The first asks for quantiles at levels 0.1, 0.5, and 0.9, while the second asks for quantiles at 0.1, 0.2, ..., 0.9. A test submission file with correct ordering fails because its quantiles are re-ordered to the levels [0.1, 0.5, 0.9, 0.2, 0.3, 0.4, 0.6, 0.7, 0.8]

[elray1]

I thought that a solution might be to update order_output_type_ids as follows, essentially doing the join/reordering within each task group:

order_output_type_ids <- function(tbl, config, types = c("cdf", "quantile")) {
  # step 1: create a lookup table from the config
  cdf_and_quantile <- config$output_type %in% types
  order_ref <- config[cdf_and_quantile, , drop = FALSE]
  # step 2: join
  tbl$output_type_id <- as.character(tbl$output_type_id)
  tbl_cols <- colnames(tbl)
  join_cols <- tbl_cols[tbl_cols != "value"]
  dplyr::inner_join(order_ref, tbl, by = join_cols)
}

However, this led to errors like:

Error in `dplyr::inner_join(order_ref, tbl, by = join_cols)`: Can't join `x$origin_date` with `y$origin_date` due to incompatible types.
i `x$origin_date` is a <character>.
i `y$origin_date` is a <date>.

So I guess we would need to update the tbl data types to the expected hub data types to enable that join... which should be OK as long as check_tbl_col_types ran previously and passed?

[zkamvar]

made changes in c8b9e67 after discussion with @elray1