I think its is pretty clear that the crew shut down the wrong engine, #1.
But what happened to ENG #2?
Unless there are some timing issues (Sample rate) on the DFDR traces released it doesn't make any sense.
The triggering condition for ATPCS is one TQ < 18%. But according to the DFDR the engine was already feathered by the time TQ #2 dropped below 18%, that sould rule out an standard ATPCS sequence, unless there are timing issues with DFDR data.
The only uncommanded feathering of a propeller on an ATR I have head about involved the incorrect installation of the prop itself, in which the mechanics omitted installing the stepped pin, which after a few flights resulting in loss of oil pressure to the prop which then by design will feather.
But that would initially lead to a massive over torque on affected engine, and even if the DFDR data does indicate an initial TQ 2 increase, its very hard to tell from the low res scale provided.