Good point. Roselawn was a hinge moment reversal that gave aileron snatch. The Indon event, it isn't noted in the report but is in the dataset, and shows a point were the wing became distorted from the aileron moment, and the aircraft rolled in the opposite direction. At that point it was still in one piece, not long after it wasn't.
The hinge moment is AOA dependent, and the ATR didn't much like the AOA that it was forced to. The pressure distribution on the wing in general from runback doesn't help at all, and does impact the effectiveness of the TE down aileron. The moment change however drove the aileron to command a right roll, as AOA increased through 5 degrees. During the recovery, they appeared to try to keep the recovery rate within the range that they had roll authority, and ran out of air. The certification requirement is to ensure that the control of the aircraft remains normal sense after the stall, but the icing went beyond the expected conditions to be encountered. At a low speed stall with manual, reversible controls the dynamic load is not normally going to be a nuisance. At higher speeds, the force can become problematic. High speed stalls are self limiting, as the stall occurs the force required to enter the stall goes away, but a few tails that have been torn up indicate if you try really really hard you can hold into buffet for long enough to break things.
Re the Saab, I'll PM you a dropbox link... have the report somewhere around. Take home is some of the stuff around the plane has good reasons to be there.
The only vice i can think of with the SAAB was the tail stall possibility on the A model. It was fixed on the A by locking out flap 35. The B model which are by far the majority have a completely different horizontal stab, larger and no stall issue. Some A models also have a slightly different main wing, the fairchild wing.
WRT the ATR, my fault using the wrong terminology, it was a hinge moment. The airflow over the wing at specific AoA would force the aileron to move with significant force. This would lead to a strong uncommanded roll until AoA was reduced. It was not stall related, and happened at high speed as well. The DGAC got into strife because they had known about the hinge moment since certification, however rollover events in service had continually be called pilot error, until Roselawn...