rebased on master

I've added some benchmarks to the branch. For longer strings, I'm getting fromNonEmptyCharArray taking an average of 50.49 μs, and fromFoldable1 taking 69.56 μs: there's definitely a decent improvement there, though nothing order-of-magnitude.

It will be interesting to compare toNonEmptyCharArray to an eventual toUnfoldable1. See purescript/purescript-unfoldable#19

That being said, there is probably something to be said in favor of the API for NonEmptyString mirroring that of plain String. Besides which, this is almost always the conversion you want, anyway: how often will we see people converting between Strings and NonEmptyList Chars, for example?

I think this is worth adopting on those grounds.

Hey looks neat :) I would like to see CodePoint conversions as well, since they make working with unicode nicer.

👍 on codepoints, but we can add that separately as it's non-breaking at least 🙂