Huh. That's interesting but in practice it doesn't quite work. The major streaming platforms want to handle things programmatically in chunks and they need a way to establish what codec (among various other parameters) to use before they get started. So the requirement is a browser mechanism to make that information available to server provided code running on the client. And I'm further stipulating that this mechanism should facilitate optional intervention by the user.
We're talking about fingerprinting, not serving up content. They could leave the content being served up unchanged, then add those tags elsewhere on the page where the user wouldn't notice, pointing at a tiny 1-second empty video.
