tbl_summary() update for ordered factors and other class handling

[ddsjoberg]

tbl_summary() accepts ordered factors, but in a hacky way: we replace the ordered factor with a non-ordered factor.

  # removing ordered class from factor variables -------------------------------
  data <- dplyr::mutate_if(data,
                           ~inherits(., "ordered") && inherits(., "factor"),
                           ~factor(., levels = attr(., "levels"), ordered = FALSE))

The downside to this approach is that the label is stripped from the column, and it's not clean code.

I would eventually like to update so that all data types are accepted, including ordered factors. The following steps would need to be taken.

• Investigate the location int he code base that erred when ordered factors were passed, and update.
• Update to accept base R date and datetimes.
• The current reason we do not accept non-base R classes, is because we don't know how to summarize them.
  • For non-base R classes, should we print an error, and require the user to specify the summary type (continuous, categorical, dichotomous)? Perhaps, for classes that we don't know how to summarize, we can omit them from the table and print a note that we didn't know how to summarize the non-base R class. Then direct them to specify the summary type OR convert to a base R class to have it included in the table.
  • Should we just guess and send the column through the defaulted summary functions? If we do this and there is an error, we should print an informative error message indicating the non-base R class passed, AND the function used to summarize. Ask the user if this class is supported in the summary function.

[ddsjoberg]

This update will have implications for #488