Add a 'remote' property to MediaSourceInterface. Also adding an implementation to the relevant sour…

talk/app/webrtc/rtpsender.cc

Index: talk/app/webrtc/rtpsender.cc
diff --git a/talk/app/webrtc/rtpsender.cc b/talk/app/webrtc/rtpsender.cc
index ea10b7b33f9b4e6ef3f92c8e5eab6124b2ac816b..c0d23a0503cc08fb5a7fd8b301f8bd99bf076bd8 100644
--- a/talk/app/webrtc/rtpsender.cc
+++ b/talk/app/webrtc/rtpsender.cc
@@ -184,7 +184,8 @@ void AudioRtpSender::Stop() {
 void AudioRtpSender::SetAudioSend() {
   RTC_DCHECK(!stopped_ && can_send_track());
   cricket::AudioOptions options;
-  if (track_->enabled() && track_->GetSource()) {
+  if (track_->enabled() && track_->GetSource() &&
+      !track_->GetSource()->remote()) {
     // TODO(xians): Remove this static_cast since we should be able to connect

[perkj_webrtc, 2015/12/15 10:15:31]

Should we rephrase this todo to say that we should look into handling
cricket::AudioOptions options from remote audio sources (whatever that means in
the end. With a little luck there will not be any constraints/AudioOptions
needed in a track/source that a PeerConnection must care about.

[tommi, 2015/12/15 11:00:54]

I'd like to keep the TODO untouched for now so that it's easy to get to the CL
that added it from the blamelist and see the context.  Right now I think that
this isn't needed at all and is from a time when the APM was inside
PeerConnection and not on a local track.  With a little luck I'll delete this in
the next few days.

     // a remote audio track to a peer connection.
     options = static_cast<LocalAudioSource*>(track_->GetSource())->options();