update the way we glob team files

[ashleywillard]

@martinemde says that this change will make these globs a bit faster which is important when you're running this on a large codebase.

[technicalpickles]

this change will make these globs a bit faster

The main reason I'd expect this is that it only tries recursing through the code base once for owned and once for unowned w/ all the glob patterns, as opposed to recursing once per glob.

It might be possible to improve further by doing one giant glob, and using File#fnmatch? to exclude unowned patterns. benchmark-ips + benchmark-memory is super helpful for comparing multiple implementations

[martinemde]

I went with a slightly different Dir.glob(code_team.owned_globs) - Dir.glob(code_team.unowned_globs) that is used elsewhere for now. It might be worth revisiting for fnmatch, but I'll do it in another commit.

[martinemde]

Eh, I couldn't help it. The ugly part here is that unowned_globs is an array, so at best we have to do the following which, thought it might be faster, looping that for every file just annoys me 😬

unowned_globs.any? { File.fnmatch?(_1, filename) }