I would strongly suggest against forcing users to produce, this should always be optional when joining session. Handling the two types of stream a separate transport for both the audio/video. So if we have webcam+mic and screenchare + audio that’d be two transports and for each viewer another two (each).
With some platforms like iOS, there’s the need to request audio/video persmissions through GUM atleast once to be able to auto-play each video when you subscribe. So for that, I would say before user joins room to do a permission check of their devices quickly whether they use the devices or not and explain to user this is required for auto-play.