Case 1: how well balanced is the stick in these aircraft to longitudinal accel?
If not perfectly balanced, the effect of the accel on the stick is to generate an apparent inertial force AFT (think of the stick as an upside-down pendulum, with most of the weight at the top of the stick: as the aircraft (pendulum pivot) accels either you have to push on the top of the stick to keep it vertical OR it moves back under the 'g'). That might cause out-of-trim behaviour during accel.
I know for certain that the BAe Hawk isn't balanced longitudinally and had to be specifically rebalanced as the T-45 due to the much higher accels on the cat (Hawk regular accel isn't big enough to matter). At 1'g;'-ish accels with lightweight fighters in AB, maybe it matters? Just a thought.
And for all three cases, unless the FCS is predictive, it can't counteract a roll, say, until it sees a roll. So once the bomb drops, the FCS won't (I suspect) apply roll control until the disturbance has started. It's technically impossible to completely zero out the controlled parameter in a simple feedback loop.