henra:
Here comes my scenario which I have in mind based on what we know atm :
2:09:30 Cb with supercooled droplets. Pitots starting to freeze over. 2:10:00 - 2:10:15 Pitot fail, unreliable airspeed. 2:10:15 - 2:10:30 AP A/THR disconnect. (Ovspd induces thrust redux, decel) 2:10:30 - 2:10:45 ADR disagree, Alternate Law 2 2:10:45 - 2:11:30 (significant control inputs not covered by protections => overcontrol => accel stall => wingdrop=> loss of spatial orientation and forced to Unusual Attitude Recovery on instrument scan after stall/upset
2:11:30 - 2:15:00 Aircraft progressively stalled.
henra, my only comment is your supposition of the nose low attitude (20-30 deg nose low) ... which I think you intend to be attitude at impact? (If not, then I misread your post).
I see the crew confronted with three different challenges all at once
1. The stall/upset (agree with your posited wing drop)
2. Stall recovery (3-D) using an IFR scan
3. Unusual attitude recovery on an IFR scan directly related to the stall recovery
I'll add to numbers 2 and 3 (caveat, it may be a load of rubbish if the gyros in the A330 are not prone to tumbling ...)
"these recoveries to be accomplished on instruments with an attitude gyro that has tumbled." (As I was speculating about some posts back.) In other words, a spin (or at least a rotating sort of stall) followed by an unusual attitude recovery, all with a partial panel scan.
There would be an unknown interval of time before the crew recognized that their primary attitude reference is buggered (possibly shown by the PFD reset attempts?) and they are called upon to recover from a 3-D upset using a partial panel scan. Is this trained for? (From my own experience, that particular taks is hard, even when you are ready for it, and in a training environment).
Added to this degree of difficulty is a reasoned belief that for a good portion of the descent, airspeed indications, (primary ref on partial panel scans) are unreliable.
If all I had to rely on to recover from an under the bag spin recovery was turn needle, balance ball, and Vertical speed, I might or might not manage it on the first go. I was trained to use airspeed as a primary reference for partial panel scan.
The aircraft gets to a high RoD ... then airspeed indication finally come back (at lower alt) and the pilots begin to catch up ... maybe unstall again, and then get stuck in a problem of accelerated stall at the end of it all as they 'pull up' (altitude keeps getting smaller in large chunks, one has to break the descent ...). Given "in the goo" scan and possibly disoriented pilots, the plane is in the process of a major pull up (nose at or above horizon) when it restalls yet again, accelerated, and falls ... this time, with no altitude left. (Hence not quite hitting in the nose low attitude you suggest ... )
I base this amplified version of your scenario on a mishap that killed a colleague of mine. He was in a spin or a spiral, went into clouds, and as he came out (finally not in the clouds anymore) was making a major correction to his high rate of descent when he got into an accelerated stall "close to the ground." The eye-witness saw the vapor trails coming off the wings (similar to what you see the Thunderbirds or Blue Angels create in their high G turns at airshows) and reported that he saw the aircraft abruptly controlled flight with little to no altitude left for one last recovery ...
That was a lot of text to suggest why the BEA nose attitude assessment might be spot on, within the general scenario you presented.