Geo: store clone URLs in GeoNode for the GeoLogCursor replication

When we move away from SystemHooks to GeoLogCursor replication (#2467 (closed)), we may loose the clone URLs information that we used from the hook. There is no sane way to figure out the URL as this is a configuration in gitlab.rb file and is not persisted on the database.

The idea is to have an initializer that will update GeoNode with current clone URL initial part, so we can build URLs from the secondary node.