Flip unified media pipeline to default-on w/ disabled holdback.

content/browser/gpu/gpu_process_host.cc

Index: content/browser/gpu/gpu_process_host.cc
diff --git a/content/browser/gpu/gpu_process_host.cc b/content/browser/gpu/gpu_process_host.cc
index cf45dab533762492376bc9b587848dccb59cc82d..c16a63968238191fc66a1ee016bd165350eb6e6c 100644
--- a/content/browser/gpu/gpu_process_host.cc
+++ b/content/browser/gpu/gpu_process_host.cc
@@ -108,9 +108,6 @@ static const char* const kSwitchNames[] = {
 #endif
   switches::kEnableHeapProfiling,
   switches::kEnableLogging,

[ncarter (slow), 2016/03/30 17:39:55]

If the old code required kEnableUnifiedMediaPipeline to be propagated to the gpu
process, why doesn't the kDisable switch need to be propagated?

[DaleCurtis, 2016/03/30 17:46:21]

This was left over from old requirements. Nothing in the GPU process uses this
switch anymore. We used to use it when choosing which rendering strategy to use.
We now always use the zero-copy path except when WebView is enabled.

-#if defined(OS_ANDROID)
-  switches::kEnableUnifiedMediaPipeline,
-#endif
 #if defined(OS_CHROMEOS)
   switches::kDisableVaapiAcceleratedVideoEncode,
 #endif