Release media session and power save blockers for casting.

media/blink/webmediaplayer_impl.cc

Index: media/blink/webmediaplayer_impl.cc
diff --git a/media/blink/webmediaplayer_impl.cc b/media/blink/webmediaplayer_impl.cc
index 64a44f770d4990fecbb35f8ce4b99328774cdd9f..df41ab2c371504722ad4d0c11466756bd247cafe 100644
--- a/media/blink/webmediaplayer_impl.cc
+++ b/media/blink/webmediaplayer_impl.cc
@@ -883,6 +883,8 @@ void WebMediaPlayerImpl::OnPipelineSeeked(bool time_updated) {
 void WebMediaPlayerImpl::OnPipelineSuspended() {
 #if defined(OS_ANDROID)
   if (isRemote()) {
+    if (delegate_)
+      delegate_->PlayerGone(delegate_id_);

[hubbe, 2016/03/18 21:40:24]

This doesn't seem right.
The browser side already treats playing/pausing differently when is_remote is
set. Or am I missing something?

[DaleCurtis, 2016/03/18 21:43:38]

AFAICT WMPI::pause() is never called when suspending for remote. So I turn it
back to you, am I missing something? :)

[hubbe, 2016/03/18 23:09:43]

I could have sworn this was working correctly at some point.
I don't suppose we have a test for this somewhere to make sure we don't screw it
up again?

Calling PlayerGone seems really weird to me, is that really the right function?
Do we need to do anything when we come back from remoting?

[DaleCurtis, 2016/03/18 23:22:24]

OnPipelineResumed() will restart playback if necessary and send the right
signals. You do want PlayerGone() here since we want the media session to
disappear. I haven't tested since I don't have a cast setup here, but if you
want to verify that this is necessary that'd be really helpful.

     scoped_refptr<VideoFrame> frame = cast_impl_.GetCastingBanner();
     if (frame) {
       compositor_->PaintFrameUsingOldRenderingPath(frame);