The pilots could over-rule MCAS at any time. Press the trim switch on the wheel and MCAS trim command stops. That's clearly seen in the FDR graphs. The pilots retained the ability to set whatever trim status they wanted at any time. The source of the problem is the SMYD with a side of ADIRU - their design to pass on false AoA and a false stall warning is the source of the problem. As I mentioned even the Autopilot was trying to force the nose down when the ET302 crew got it to engage. Forcing the nose down on a stall warning is the action pilots should take (though they should actually set pitch and power appropriate for the weight and altitude.)
They changed the switches because (AFIAK) the NG guidance was that for any trim runaway, both switches are to be used, so operationally it made no difference.
The first Lion Air crew flew just fine with the manual trim wheel. It went so fine, they re-enabled electric trim, MCAS gave another kick, and they turned it back off.
If you are looking for the regulatory smoking gun, that was fired a long time before MAX was a twinkle in anyone's eye when they allowed a false stall warning to be acceptable and did not seek a way to invalidate a failed AoA sensor, perhaps as detected by loss of continuity in the de-ice circuit which occurs when the external vane is stripped from the plane.