Inform jitter buffer about FlexFEC protection.

webrtc/call/call.h

Index: webrtc/call/call.h
diff --git a/webrtc/call/call.h b/webrtc/call/call.h
index fdb6cff785b7e083be2517a5dbf4b088544abfbb..cb23605f4b6bffa16a3d824cbad35417f39a9bb0 100644
--- a/webrtc/call/call.h
+++ b/webrtc/call/call.h
@@ -119,6 +119,9 @@ class Call {
   virtual void DestroyVideoReceiveStream(
       VideoReceiveStream* receive_stream) = 0;
 
+  // In order for a created VideoReceiveStream to be aware that it is
+  // protected by a FlexfecReceiveStream, the latter should be created before
+  // the former is started.

[brandtr, 2017/01/24 13:02:49]

An alternative design would be to make VideoReceiveStream implement an interface
"FlexfecProtectionObserver" that would be called by Call whenever a new
FlexfecReceiveStream is created that protects that particular
VideoReceiveStream.

That design would however require either recreating the receiver objects (e.g.,
by calling Stop() and Start()), or creating a new "method path" down to the JB
that propagates the information that FEC has been dynamically enabled. Neither
seem like great choices, so I went with this sort of static design instead. The
downside of this is that we must always create the FlexfecReceiveStream before
Starting the VideoReceiveStream. There aren't too many places where the
VideoReceiveStream is started(*), so the chance of making mistakes is not big
however.

(*) VideoReceiveStreams are started in
WebRtcVideoChannel2::WebRtcVideoReceiveStream::RecreateWebRtcStream (production
code), CallTest (E2E and VSS tests) and VideoQualityTest (perf tests).

   virtual FlexfecReceiveStream* CreateFlexfecReceiveStream(
       const FlexfecReceiveStream::Config& config) = 0;
   virtual void DestroyFlexfecReceiveStream(