Move direct use of VideoCapturer::VideoAdapter to VideoSinkWants.

webrtc/media/base/videosourceinterface.h

Index: webrtc/media/base/videosourceinterface.h
diff --git a/webrtc/media/base/videosourceinterface.h b/webrtc/media/base/videosourceinterface.h
index 14624585432194c74bec680c6f18befb56ce79f5..cfe812319c91a77a4b7644f2a44500b730d4cbc4 100644
--- a/webrtc/media/base/videosourceinterface.h
+++ b/webrtc/media/base/videosourceinterface.h
@@ -13,6 +13,7 @@
 
 #include "webrtc/media/base/videosinkinterface.h"
 #include "webrtc/base/callback.h"
+#include "webrtc/base/optional.h"
 
 namespace rtc {
 
@@ -27,6 +28,12 @@ struct VideoSinkWants {
   // Tells the source whether the sink wants frames with rotation applied.
   // By default, the rotation is applied by the source.
   bool rotation_applied = true;
+
+  // Tells the source the maximum number of pixels the sink wants.
+  rtc::Optional<int> max_pixel_count;
+  // The sink can consume this amount of pixels but wants more. The source
+  // should produce a resolution one "step" higher than this.

[pthatcher1, 2016/02/25 07:40:30]

Can we add a little to this?  Perhaps like so: "Like max_pixel_count, but
relative to the given value.  The max should be one "step up" from the given
value.  In practice, this means the sink can consume this amount of pixels but
wants more and the should produce a resolution one "step" higher than this, but
not higher".

[perkj_webrtc, 2016/02/25 13:57:07]

Done.

+  rtc::Optional<int> max_pixel_count_step_up;
 };
 
 template <typename VideoFrameT>