width/height metadata on incoming streams

[fippo]

The video frame metadata come with width and height:
https://w3c.github.io/webrtc-encoded-transform/#dom-rtcencodedvideoframemetadata-width

For outgoing streams this makes some sense since it is easily known (but it isn't clear to me what the use-case is), for incoming streams this requires pre-parsing the VPx payload descriptor, the H264 SPS (or PPS?) and the AV1 sequence header OBU.

Is anyone relying on this behavior? Note that it already doesn't work for AV1 so the values are 0/0 and I heard a "we would like to move away from parsing bitstreams" here that I agree with.

While I don't think we can change this to be optional depending on the direction, we can at least consistently set width/height to 0 on incoming streams.

[alvestrand]

These are dictionary members without "required" on them, so omitting them is completely consistent with syntax.
I would much prefer "missing member" over 0/0 as "there's nothing here to see" indication.