Remove/replace runtime dependence on more-itertools, jaraco.functools, and jaraco.collections

[Sachaa-Thanasius]

A very small diff that would remove 3/4 of distutil's current dependencies.

Imo, this makes distutils easier to understand by replacing opaque helpers with a) well-known stdlib alternatives, or b) obvious and short idiomatic Python. It also should make vendoring distutils easier (e.g. what setuptools does).

(Re. tests: I ran them locally on Windows via tox, and the results before and after were the same — everything passed besides 1 unrelated cygwin thing.)

(I can make an issue about this first if need be — the diff is so small that I figured this wouldn't hurt, though.)

[effigies]

Hope drive-by reviews are welcome. This makes sense to me, and I have one small suggestion.

[Avasam]

I know Jason is a strong proponent of DRY, even if it means extra simple dependencies (there's a reason there's so many jaraco packages 😝 ). So I'd expect some push back.
Aside from whether the changes will be accepted in the first place, the new code looks good to me, as a non-maintainer.

[Sachaa-Thanasius]

I know Jason is a strong proponent of DRY . . . . So I'd expect some push back.

Yeah, I was hesitant to open this PR because of that, but figured it was worth a shot. Drastically reducing the dependency "surface area" at the cost of a tiny bit of repetition (and really, it's quite tiny in this case) felt like a reasonable tradeoff.

(And that repetition, if it even qualifies as repetition, is just for the code replacing jaraco.functools and more-itertools - the replacement for jaraco.collections.DictStack doesn't repeat at all)