Franzl - it looks to me (quickly doing the maths in my head) like the elevator trace looks like it takes THS position into account and is relative to the THS value.
The FCPC will command what it is told to, this is true - but letting go of the stick should at least allow the aircraft to follow natural trajectory - it wouldn't be enough to get the nose down, and I'm not arguing that it would, but the elevators were being held in position by the inputs and came back down when the stick was held forward for a few seconds and then held around neutral. The question of whether the underlying systems were in G-loading mode or pitch command mode (which IIRC becomes active below a certain speed) is important when discussing this, because if it was holding G then the FCPC might have held the elevators there and if it was in pitch command mode the elevators should have relaxed.
Either way this doesn't help understanding to a great degree, because at the point the elevators returned from full-up, the aircraft was beyond saving. Why did it take so long?