use libgit2 for git operations to improve performance and security

[MoralCode]

Libgit2 (website, repo) is a cross platform C library implementing the git plumbing. It is designed for use as a library within complex applications, unlike the git CLI which is designed as a CLI. (source)

https://www.pygit2.org/ is the python version of this.

I learned about libgit2 through a conversation in a nerdy group chat of mine where there was a large 80ms -> 0.5ms speedup for one minimal usecase (render times for displaying git repo status in your shell prompt). While this may not be applicable to Augur directly, I wonder if we could eliminate a lot of shell calls, parsing bugs, and extra layers in the process when facade collects a git repo by using git more directly via a library that directly interfaces with git's plumbing, rather than calling an external program to do it.

[MoralCode]

this is more of an optimization, but might help fairly significantly with the performance of processing large repos through facade (since we arent bound by API limits for the most part).

I think it would be nice to do an experiment, such as copying augurs facade parsing code to a standalone script to directly benchmark the two implementations against each other in some small and large repos just how much there is to gain here.

I feel like it could be a lot as this would eliminate the overhead of spinning up a subprocess, starting an external executable, and then reading and processing its output line by line. Replacing it with a proper library implementation of git that has been around since 2014 (IIRC) would mean gettting the information we need could be as simple as a function call and would be a much more direct path.