HarryMann,
That was a most thoughtful and relevant post.
Just a few thoughts of my own, based first on your numbered ones:
1) The inhibition of the stall-warning when the sensed airspeed falls below a certain value needs to be addressed. I suspect the problem is that the present generation of AoA sensors cannot operate accurately at very low TAS, which is what they experience every time an aircraft vacates the runway after landing. Ground/flight detection remains, unfortunately, less than 100% reliable. That may sound ridiculous in the context of AF447, but it's part of the problem for the designers.
2) Agree as far as configuration is concerned. There is no way that you would ever want a THS setting of 13NU on a clean aeroplane. Even half that value is a frightening thought.
3a) Rightly or wrongly, this is standard industry practice: not an Airbus-ism.
3b) The constraint is an attempt to avoid false stall warnings, which themselves are dangerous, but it failed the crew in this instance.
Re some of your other points.
The captain had an impossible task on his return to the cockpit. In any case, the "ride" would have made any activity/observation extremely difficult, unless and until he managed to get into the P3 or P4 seat and attach seat-belt or full harness. (The view from the P4 seat is likely to be poor.)
Regarding your other comments, I think the emphasis on AoA would involve revolutionary changes in civil pilot training; starting ab-initio. To justify that, we would need to consider how much a lack of AoA awareness may have contributed to other accidents, not only the obvious ones like G-ARPI, which also involved an unawareness of wing configuration.
In the case of AF447 the bottom line remains, however: why did the PF demonstrably embark upon and maintain a clearly unsustainable climb from level flight?