Figure out layer for services like Thumbnails

[clone1018]

PeerSnap works by initiating a WebRTC Peer Connection with the WHEP endpoint that Waveguide presents, and "watching" the stream by consuming it's RTP packets until enough have been collected to produce a thumbnail of the video.

The benefit of doing this type of work on the "output" is that we've already messaged the stream into RTP packets, and WebRTC is doing it's magic to codify that, handle dropped packets, and otherwise deliver a reliable stream.

This type of service is likely going to be common in Waveguide, to provide abstract ways of collecting thumbnails, metadata, and other functions that all inputs would otherwise need to implement.