Lastly, as I understand it, MCAS only dials in *ONE* nose-down increment (2 degrees or whatever it is) and then deactivates itself, until something happens to reactivate MCAS allowing it to dial in another increment. I can't recall all of the crew activities that result in MCAS being reactivated, but I believe one is operation of the manual trim. This is why the last four "blipped" manual trim inputs in the Lion flight result in four unmitigated MCAS nose-down events, because any manual trim input resets MCAS and allows it to reasses the AoA / speed picture and dial in another MCAS trim input.
MCAS is perpetually active so as long as ONE AoA sensor relays: high angle of attack, flaps up and autopilot disengaged. Simply, it will trim the Stabilizer down for 10 seconds (2.5 degrees nose down) and pause for 5 seconds before doing it all over again if it deems conditions remain the same (which often does with faulty sensor data; in the case of MCAS it's engaged on single channel FCC). You could have an aircraft ~4 degrees nose down from level flight as early as 45 seconds.
Application of electric trim pauses MCAS (5 seconds). Selecting TRIM CUTOUT to OFF deactivates MCAS (much like Speed Trim, Mach Trim and Column switch electric trim)
So if they were manually trimming the stab wheel as you say, one would assume they'd have tried either of those two options which begs the question why wouldn't they have been able to control the aircraft in the pitch axis thereafter? Time will tell.