Add findnearest_partial and findall_partial

[ericphanson]

This PR adds two functions that act like Partial except give indices of the match (or of all matches) in addition to the closest distance. I've found this useful for fuzzy matching keywords in long documents (via KeywordSearch.jl, which adds some types and more API on top of these functions). Having the indices in that context is important since it lets you quickly check if the match looks genuine or spurious (i.e. does it look like a typo, or is it a completely different word). To get the indices of the match, we need the collection to be indexable, not just iterable. That's more restrictive than the rest of StringDistances but I think it's necessary for this functionality to work.

Let me know what you think; this code can stay in KeywordSearch if you prefer, but I thought it might be more useful here.

Closes #29 which was an initial version of this.

[coveralls]

Coverage increased (+0.2%) to 94.757% when pulling efa3553 on beacon-biosignals:eph/find_partial_match_indices into d650b62 on matthieugomez:master.

[ericphanson]

Bump

[ericphanson]

Any thoughts on this?

[ericphanson]

Bump 🙂 . Found another use for this! Would be great to get it in here.

[matthieugomez]

Thanks for the PR. I think it would be better in another package, I’m not sure this is a very common operation.

[ericphanson]

Ok, makes sense. It'll be in KeywordSearch.jl if anyone needs it 🙂.