Why not do both? Use rsync to do a one way sync by keeping a filter list to the paperless consume directory so whenever you drop any new documents to your original folder structures, it copies THAT to the consume and paperless consumes it?