Locking and transactional/atomic updates for config

This makes the API for commiting or discarding changes the same as for
references.

This lock/unlock pair allows for the cller to lock a configuration file
to avoid concurrent operations.

It also allows for a transactional approach to updating a configuration
file. If multiple updates must be made atomically, they can be done
while the config is locked.

When a configuration file is locked, any updates made to it will be done
to the in-memory copy of the file. This allows for multiple updates to
happen while we hold the lock, preventing races during complex
config-file manipulation.

While we download the remote's remote-tracking branches, we don't
download the tag. This points to the tag auto-follow rules interfering
with the refspec.

When an error state is an OOM, make sure that we treat is specially
and do not try to free it.

 filebuf: remove lockfile upon rename errors

When we pass the path of a repository to `_bypath()`, we should behave
like git and stage it as a `_COMMIT` regardless of whether it is
registered a a submodule.