If frames were delivered within the same millisecond as VideoCaptureImpl
was created, or the timestamp weren't granular enough then the first
frame would be mistakenly dropped because of having the same timestamp
as a previous one, even though there was no previous one.
BUG=
R=mflodman@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2599004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@5004 4adac7df-926f-26a2-2b94-8c16560cd09d
There's no real notion of VideoEngine left in these classes. They're
end-to-end tests built on Call, so CallTest makes more sense.
This also contains a modification to RtpRtcpObserver moving the
responsibility of creating the event that signals when the observation
is complete to RtpRtcpObserver. New tests are about to be introduced and
this will reduce code duplication.
BUG=
R=stefan@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/2258005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4793 4adac7df-926f-26a2-2b94-8c16560cd09d
