Improve some error messages

Some potential misleading error messages :

- If there are no unsafe cells, a message to this effect is returned by tab_rtauargus, but also an error.
- Error in `[.data.frame`(tabular, , c(explanatory_vars, value, freq, maxscore, : undefined columns selected – this error occurred with amount tables and went away when I commented out cost_var = “n_obs”. Oddly, it worked w/o error when I misspecified cost_var= “n_obs” as cost = “n_obs”.
- Error: cannot allocate vector of size 361.8 Mb - list_explanatory_vars was missing its names
- Error in if (sum(has_primary_secret) == 0) { : missing value where TRUE/FALSE needed - there was an additivity problem in the data. I have checks for this now.
- Error in `map()`: … Cause by error in `[[<-.data.frame`: ! replacement has 0 rows, data has 28728 – duplicate name in list_tables so mismatch with list_explanatory_vars
- This one occurred at , however, the inputs were fine, the TA inputs created by rtauargus were fine, but there was some issue with the file permissions of existing files. Error in read.table(file = file, header = header, sep = sep, quote = quote, : more columns than column names.

Start by reproducing the listed failures through tab_rtauargus and the path, including the no-unsafe-cells, amount-table, missing names, additivity, duplicate-name, and file-permission cases. Trace where each underlying error is surfaced and document the expected actionable message for each condition; done means the misleading errors are replaced without hiding genuine failures.

Written by the indexing model from the issue text.