Faulty AoA information
Hmm. So, my understanding of what FCeng84 said is that MCAS is essentially a 'one-shot' pitch ND command when AoA is determined to be 'high'; pulling back on the yoke does not re-arm the MCAS trigger, but activating the trim switches or manually spinning the trim wheels does re-arm the trigger; and if the AoA drops below the threshold MCAS is re-armed.
One other thing occurs to me: if the MCAS system is receiving faulty AoA information, it is not beyond the bounds of possibility that the AoA information received varies, so it is possible that the reported, erroneous, AoA could rise to a value high-enough to trigger MCAS, drop enough to re-arm the MCAS trigger, then rise again to trigger the MCAS action, and do this repeatedly. Of course, there may be some logic to detect this, but in normal flight, nobody expects AoA to see-saw through the trigger threshold repeatedly.
I don't know if the traces for AoA and Automatic Trim from the FDR are capable of showing whether or not such a behaviour occurred. You'd probably need to know the fine details of how the MCAS threshold varied according to reported AoA and Mach number, and I'm not in a position to determine that.
Semreh