cc: Add TaskGraphRunner sub-namespaces with task concurrency limits.

cc/resources/gpu_tile_task_worker_pool.cc

Index: cc/resources/gpu_tile_task_worker_pool.cc
diff --git a/cc/resources/gpu_tile_task_worker_pool.cc b/cc/resources/gpu_tile_task_worker_pool.cc
index c57d25a27de533214ba2656c00bbfad32d2f5553..469987a41b91ce675806637569b68bf84ceca1c2 100644
--- a/cc/resources/gpu_tile_task_worker_pool.cc
+++ b/cc/resources/gpu_tile_task_worker_pool.cc
@@ -145,13 +145,15 @@ void GpuTileTaskWorkerPool::ScheduleTasks(TileTaskQueue* queue) {
           TaskGraph::Edge(task, new_task_set_finished_tasks[task_set].get()));
     }
 
-    InsertNodesForRasterTask(&graph_, task, task->dependencies(), priority++);
+    InsertNodesForRasterTask(&graph_, task, task->dependencies(), priority++,
+                             kGpuSubNamespace, kGpuMaxConcurrentTasks);
   }
 
   for (TaskSet task_set = 0; task_set < kNumberOfTaskSets; ++task_set) {
     InsertNodeForTask(&graph_, new_task_set_finished_tasks[task_set].get(),
                       kTaskSetFinishedTaskPriorityBase + task_set,
-                      task_count[task_set]);
+                      task_count[task_set], kDefaultSubNamespace,
+                      kDefaultMaxConcurrentTasks);
   }
 
   ScheduleTasksOnOriginThread(this, &graph_);