is match faster than group_indices?

[mdsumner]

https://stat.ethz.ch/pipermail/r-help/2018-November/460447.html

[mdsumner]

fastest now is vctrs::vec_group_id and that's trivial to use with column name subset - the issue with unjoin is being general for any column choice with tidy.

[mdsumner]

Maybe: https://cran.r-project.org/web/packages/tidyselect/vignettes/tidyselect.html