A Squared,
thank you so much for having so clearly explained your oppinion on this subject.
I agree with every sentence you have written, and indeed I also use the same arguments next to those who keep finding that the excessive roll due to torque is correctly modeled on some flight simulators.
The spiraling slipstream hitting the vertical fin / rudder, which is usually above the CoG of the aircraft, as well as the down sides of left wing and left side of the horizontal stab also accounts for a moment that is opposed to that of reactive torque. Add to it the mild force that is actually produced by the engine/prop torque on your C172 example, and we have excellent reasons to understand why roll is far from being the main effect...
All summed up together will certainly account for very mild roll effects as opposed to yaw - which indeed should / is the prevalent effect.