feat: search first character of each word

[christoph-heinrich]

I don't really understand why title and hint characters should be concatenated.
It might make some sense when the title is only one word, but otherwise it might lead to unexpected results.

I tried also a different implementation based on find instead of gmatch that I expected to be faster, because it only allocated half as many strings, but it turned out to be the same speed.

Ref. #625 (comment)

[tomasklaen]

It matches non-word characters currently. For example for string "Delete file & next" it'll create a keyword df&n instead of dfn.

[tomasklaen]

And yeah, title and hint probably shouldn't be concatenated, but matched separately.

[christoph-heinrich]

We could go with a simple %w, but that doesn't work with non latin alphabets.
We could do something similar to what we're doing for word wrapping and make a list of characters that separate words.

[christoph-heinrich]

I've made a list with word separators, I hope I got everyone's favorite word separation characters in there.

[tomasklaen]

Add +&()[]{}.

And I guess put word_separators outside of the function since it's static.

[christoph-heinrich]

wrap_text also has two static lists, idk if you want to do the same thing there. If so I'll do that in the other PR that already touches wrap_text.

[tomasklaen]

Oh, and /\, sorry 😐.

wrap_text also has two static lists, idk if you want to do the same thing there

Yeah that would be nice.

[christoph-heinrich]

I missed so many characters 😞
Now that I'm writing this I realize <> should probably also be on that list.

[junguler]

amazing work guys, i really appreciate it