Geo: Automatically clean up stale lock files on Geo secondary

In a customer call today, a number of repositories were listed as failures. From geo.log, it looks like a lot of .lockfiles were older than a few days for some reason. We manually removed them via a find call.

I think we may want to issue a cleanup periodically (e.g. when a failure occurs or after a fetch) to remove these files. The is similar to https://gitlab.com/gitlab-org/gitlab-ce/issues/44115 and https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/18095, which cleans up stale lock files and worktrees before a push attempt.