Attach color space information to hardware decoded NV12 video frames.

media/gpu/vt_video_decode_accelerator_mac.cc

Index: media/gpu/vt_video_decode_accelerator_mac.cc
diff --git a/media/gpu/vt_video_decode_accelerator_mac.cc b/media/gpu/vt_video_decode_accelerator_mac.cc
index 65f00d77f5b88c1c99f00450fb9b716babb8f699..f42abd08e22cc7c9d843a93dfaf071b011c336a1 100644
--- a/media/gpu/vt_video_decode_accelerator_mac.cc
+++ b/media/gpu/vt_video_decode_accelerator_mac.cc
@@ -1105,8 +1105,10 @@ bool VTVideoDecodeAccelerator::SendFrame(const Frame& frame) {
   // GpuVideoDecoder so that GpuVideoDecoder can use correct visible size in
   // resolution changed. We should find the correct API to get the real
   // coded size and fix it.
+  // TODO(hubbe): Use the correct color space.  http://crbug.com/647725
   client_->PictureReady(Picture(picture_id, frame.bitstream_id,
-                                gfx::Rect(frame.coded_size), true));
+                                gfx::Rect(frame.coded_size), gfx::ColorSpace(),
+                                true));
   return true;
 }