This avoids falling back on the VideoFrameBuffer::CropAndScale default
implementation which performs ToI420. This has two major benefits:
1. We save CPU by not converting to I420 for NV12 frames.
2. We make is possible for simulcast encoders to use Scale() and be
able to trust that the scaled simulcast layers have the same pixel
format as the top layer, which is required by libvpx.
In order to invoke NV12Buffer::CropAndScaleFrom() without introducing a
circular dependency, nv12_buffer.[h/cc] is moved to the "video_frame"
build target.
Bug: webrtc:12595, webrtc:12469
Change-Id: I81aac5c6b3e81c49f32a7be6dc2640e6b40f7692
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/212643
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Evan Shrubsole <eshr@google.com>
Commit-Queue: Henrik Boström <hbos@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#33521}
f412976eca