Refactor with `dplyr`?

[kinto-b]

As it stands, none of the core functionality depends on dplyr. This is good, because dplyr is a fairly bulky dependency and only a fairly limited portion of it would actually be relevant/useful. The drawback is that the code I've written is slow and complicated. I could simplify the code and speed it up by refactoring using dplyr (and purrr).