Yes they must obey the laws of physics but this isn't a WISP, concurrent transmissions can't just stomp on each other and find out fast enough to hide it they've had to be broken out per client in the cell from day one. Other than mentioning the MAC layer exists is there a particular concern you have about it scaling? The satellites only have 2 paths to forward things and only a small patch of cells to do it for. Scaling has been brought by putting up orders of magnitude more satellites to provide coverage at more than the dense northern coverage from the orbits, it has not been by putting a ton more clients through the satellites themselves.
Is RF and framing something Starlink has had to design for? Absolutely, they do after all need these things. That doesn't mean it makes sense to jump to it as the source of any perceived issues, particularly on the basis it's something unknown.
And there are definitely reasonable scaling concerns here. They're using a finite set of frequencies and the laws of physics still apply. Any concurrent transmissions are going to be impacted by both their phy layer modulation strategy, in addition to the mac layer. But we know so little about those layers currently that its hard to comment without being entirely speculative.