what would have prevented the aircraft from rolling over onto its back, particularly since it had a significant tendency to roll to the right in its ALT2B mode?
Data appears to favor pilot induced oscillations and as often stated on this thread "making mayonaise".
http://www.bea.aero/fr/enquetes/vol....gure.64.en.png
Roll attitude: the pink line in chart above shows no roll over - only constant roll approx 10-15 deg. Now compare pink to red and blue and you see oscillations appear to be mostly pilot induced. I suggest that the airbus damps the roll disturbances very well on its own probably because of its mass and drag in the roll axis.