Delete incorrect comment regarding as_tibble

[brookslogan]

We say that as_tibble de-aliases data.table columns here, but it doesn't. We should still probably be good since [.data.table with an i arg (but not unique.data.table, probably a bug) seems to consistently copy. It's just the comment that's wrong.

[brookslogan]

If we wanted to be more careful here, we could try to detect column-aliasing & copy in that case.

[brookslogan]

Another probably-inaccurate comment:

  # We don't want to directly use `dplyr::mutate` on the `$DT`, as:
  # - this likely copies the entire table

[brookslogan]

It might actually be okay to alias in these cases if we never mutate archive DT columns without copying first (and if the user doesn't mutate the epi_df columns, which seems okay to assume since that's the general rule in R). I'm not sure whether that's the case or not though. It seems plausible, since IIRC we don't really do partial column modifications, though there are still key-ing operations or whole-column operations that still mutate that might be of concern.