Trigger cleanup quickly after create_distributed_table_concurrently

Currently it takes up to one citus.defer_shard_delete_interval until the Citus local table shard is dropped after a create_distributed_table_concurrently, which is confusing when doing a \d immediately after a create_distributed_table_concurrently. It would be nice if we triggered a cleanup for deferred drop more quickly, similar to how we can trigger node metadata sync immediately from citus_update_node.