ok do I have this right?
A working scenario is - the FMC disables the autopilot because of unreliable airspeed measurement by the clogged pitot tubes. Once that is done, it should not be possible to re-engage the autopilot manually. Since that apparently did happen, the implication is that the crew re-engaged the autopilot after some initial upset, and this played right into the trap that was mentioned by the recent AD. The crew think they have recovered, and instead they have guaranteed the worst sort of stall, a flat spin basically.
So all talk about being asleep and weather and all that is just not relevant - because the problem here was systemic, even down to the over-reliance on programming instead of piloting. Ok that was a speculation.