I think we are getting into the territory of finstall if memory serves. Without the speed to ensure aileron authority the rudder causes a slip towards the dead wing.
When the side slip gets too bad, it masks the rudder, coupled with all the thrust and lift on one side, inevitably the roll is uncontrollable. VMCA2 is named because the aircraft is uncontrollable below that speed without chopping the thrust on the asymmetric outboard and changing the dynamics.
On a roller, we typically rotated below 110 kts, so chopping an engine (No 1) would induce a yaw and roll easily corrected by rudder and/or bank. No 1 feathered normally and presented minimum drag. No 2 then ran down and the prop would have NTSed, which does minimise drag without feathering the prop, but causes drag nevertheless.
So with only drag on the port wing and full blown thrust/lift on the starboard, there is not enough aileron authority to control the aircraft's roll. Finstall only exacerbates an already fraught situation.
This is all from memory, so please forgive me if it's a bit hazy, it was over 30 years ago we taught this accident, to help prevent further ones.
It's perhaps the enduring legacy of this accident that it never again occurred in the RAF.
Talking of the Canberra accidents above (along with Meteors), the irony is I lost two colleagues 10 years later on 231 OCU when a simulated asymmetric exercise went wrong at Wyton. The SOP was that tip tanks had to be empty or not fitted for this exercise (EFATO - simulated), however this aircraft had half tips (1000 lb or so) in the tips and the inertia was again, uncontrollable and the aircraft crashed after rolling to the right. It was the Stn Cdr who was transiting to Kinloss for a simulator attempting to get his ticks done for the Canberra en route.
Some lessons are SOo hard to learn.