A double engine failure, or even a double engine surge, could lead to a very nasty yaw, faster than the pilot, not necessarily instantly aware of exactly what was happening, could counter.
The designers were, right from the start, aware of this problem.
Hence, the prototypes were equipped with specific "contre automatique" (auto-rudder) computers, that would "kick in" a given rudder deflection as soon as they detected an engine failure (and twice as much in the case of a double failure).
Unfortunately... the manner of detecting an engine failure was based on pressure sensors in the engine, which proved to be notoriously unreliable.
Since the whole system was "fail-passive", in the case of a pressure sensor failure nothing happened, other than that I got the "suspect" computer dumped in my lap every time, since it was easier to swap a computer than test and swap pressure sensors....
In the end, it was always "no fault found", and the engineers had to go and test the sensors to find the failed one.
Already on the pre-production aircraft, this Rube Goldberg system was replaced by a single circuit board 'buried' in the autostab computer.
It used a lateral accelerometer to detect the abrupt yaw of a sudden engine failure or surge, and applied appropriate rudder. Look at the sudden rudder deflection 'peak' on the lateral response graph in the previous post.
Since there was no separate 'auto-rudder engage' control switch (the function was permanently active), and it was only mentioned very much in passing during training, some pilots were not even aware it existed.......
CJ