In data_scanner, only the data_usage collection is running; do we still need to handle object heal, replication, and lifecycle operations