Push detached requests at the back, so they are tested last since they will most probably finish latest

mpi/src/starpu_mpi.c

@@ -1221,7 +1221,7 @@ static void _starpu_mpi_handle_detached_request(struct _starpu_mpi_req *req)
 		/* put the submitted request into the list of pending requests
 		 * so that it can be handled by the progression mechanisms */
 		STARPU_PTHREAD_MUTEX_LOCK(&detached_requests_mutex);
-		_starpu_mpi_req_list_push_front(detached_requests, req);
+		_starpu_mpi_req_list_push_back(detached_requests, req);
 		STARPU_PTHREAD_MUTEX_UNLOCK(&detached_requests_mutex);
 
 		starpu_wake_all_blocked_workers();