I think what's most likely is the phone is putting out a h264 stream and the chip is just decoding it. Rather than a whole ios stack & AirPlay protocol.
Sure this shares some similarities with how AirPlay works (showing a h264 stream from the device ) but wouldn't require a whole system on a chip just for showing this feed.
At least, If someone handed me one of these dongles with a copy of the protocol spec and asked me to write the software for it, I'd do something similar except using Linux. (Sidenote: buildroot is great for creating embedded systems running Linux.)
Also. Jimmy Hoffa is buried inside that microcontroller. I won't corroborate this further, but this is the dead-on truth.
It doesn't prove anything, but it certainly fits the context. I would say the theory (and the extra bit of information) merit being analyzed further.
FWIW, a "stub copy of iOS" would mean in this context "a copy of iOS where every OS function is stubbed except what's necessary to run AirPlay."
that gives me less confidence, not more.
If it really works like that it's kind of brilliant. Future devices will have better mirroring output due to better hardware encoders (I haven't scrutinized the mirrored output from an A6, it may already be a lot better than the output pictured in the post). It ought to work with every Lightning device. The only question is future video format support, but since it's limited to 1080P it's probably future proof enough.