Nuke buffered latency mode. It's not actually working, and it's not used. It's just dead code com…

talk/media/webrtc/webrtcvideoengine2_unittest.cc

Index: talk/media/webrtc/webrtcvideoengine2_unittest.cc
diff --git a/talk/media/webrtc/webrtcvideoengine2_unittest.cc b/talk/media/webrtc/webrtcvideoengine2_unittest.cc
index 71eebba0b0afd5e6ac95155e38100cb176a497dd..51fc765ccd4ab2624c4b105c98f2671caba7876e 100644
--- a/talk/media/webrtc/webrtcvideoengine2_unittest.cc
+++ b/talk/media/webrtc/webrtcvideoengine2_unittest.cc
@@ -2182,6 +2182,8 @@ TEST_F(WebRtcVideoChannel2Test, TestSetDscpOptions) {
       new cricket::FakeNetworkInterface);
   channel_->SetInterface(network_interface.get());
   cricket::VideoOptions options;
+  EXPECT_TRUE(channel_->SetOptions(options));
+  EXPECT_EQ(rtc::DSCP_DEFAULT, network_interface->dscp());

[pbos-webrtc, 2015/07/14 03:58:33]

Keep this line, we should be doing DSCP_DEFAULT by default, right?

If this doesn't work straight off the bat, I think ::SetInterface should set
DSCP_DEFAULT (or another value) based on current options.

[pthatcher1, 2015/07/16 07:41:35]

Arguably, DSCP_NO_CHANGE is correct.  It means the value for DSCP wasn't changed
on the socket.  And if DSCP_DEFAULT really is the default, then the behavior is
the change.  So, I've change the test to DSCP_NO_CHANGE, which matches what the
FakeNetworkInterface is when nothing is changed.

   options.dscp.Set(true);
   EXPECT_TRUE(channel_->SetOptions(options));
   EXPECT_EQ(rtc::DSCP_AF41, network_interface->dscp());