Provide task runner to GLES2Impl / CommandBuffer at creation

services/ui/public/cpp/gles2_context.cc

Index: services/ui/public/cpp/gles2_context.cc
diff --git a/services/ui/public/cpp/gles2_context.cc b/services/ui/public/cpp/gles2_context.cc
index ba8cc3b52da86db84e863a761dc6e9cc5cfa2064..9197c7f5d6e269036aaf5839d890caf303b5a621 100644
--- a/services/ui/public/cpp/gles2_context.cc
+++ b/services/ui/public/cpp/gles2_context.cc
@@ -40,8 +40,7 @@ bool GLES2Context::Initialize(
       base::ThreadTaskRunnerHandle::Get();
   command_buffer_proxy_impl_ = gpu::CommandBufferProxyImpl::Create(
       std::move(gpu_channel_host), surface_handle, shared_command_buffer,
-      stream_id, stream_priority, attributes, active_url,
-      std::move(task_runner));
+      stream_id, stream_priority, attributes, active_url, task_runner);
   if (!command_buffer_proxy_impl_)
     return false;
   gpu::CommandBuffer* command_buffer = command_buffer_proxy_impl_.get();
@@ -63,7 +62,7 @@ bool GLES2Context::Initialize(
   implementation_.reset(new gpu::gles2::GLES2Implementation(
       gles2_helper_.get(), NULL, transfer_buffer_.get(),
       bind_generates_resource, lose_context_when_out_of_memory,
-      support_client_side_arrays, gpu_control));
+      support_client_side_arrays, gpu_control, std::move(task_runner)));
   if (!implementation_->Initialize(default_limits.start_transfer_buffer_size,
                                    default_limits.min_transfer_buffer_size,
                                    default_limits.max_transfer_buffer_size,