Bump libgit2 - #359

The only minor thing that I see is that the return type of git_clear_index has been changed to indicate whether it was successful. We should probably use that. But since that's not a regression, it'd be fine to open an issue for this and do it later.

Apart from making sure that we're up to date in our usage of the APIs, I don't see any reason not to merge this to master.

👍 Filed as #361.

The only minor thing that I see is that the return type of git_clear_index has been changed to indicate whether it was successful. We should probably use that. But since that's not a regression, it'd be fine to open an issue for this and do it later.

Just for reference, the rationale for that change is that the index now uses deferred deletes for index entries if there is a concurrent diff running using the index. In theory the bookkeeping logic for the deferred delete could have a failed memory allocation so we had to add an error return.

Since I think that y'all will not concurrently modify the index with a diff in progress on another thread, this function should never actually return an error.

@arrbee 💎 Thanks for explaining!