Add option to use_raw_dataset() to automatically list the dataset in .Rbuildignore

[gfinak]

The rationale is that large data sets could be stored in the package source tree, but would not be included in the built package. It would be up to the user whether the large data sets should appear in version control or not.
Ultimately if not, we'd like other solutions to back up those large data sets. Cloud-enabled rsync-like tools exist and may be an option.