robdean... I have commented above that this is going to be "trick cyclists" investigation
Your analysis is probably along the right lines IMHO... and deeper than most of us aeroplane types can understand
If the the above described human error did occur, the Airbus system exacerbated an error for which circumstances left little margin. That does not mean that Airbus are necessarily in error or inferior, but there is surely a case for examining whether the auto thrust / braking logic trees could be refined.
This Airbus system has been around a long time... and generally successful. Changing it introduces other risks, and what's more takes a long time (and £££ / $$$$)
I am unaware of any other "failure to close TL(s)" incidents / accidents
except with the TR Inop scenario. You have given us the psychobabble of how that drill maybe led to the next "error"... If one was to change, as Airbus have done, the first part - namely a "non-standard operation involving just 1 TL" - does this maybe solve the human factors part?
In fact, the new Airbus procedure says to select both TLs to Full Reverse. This also applies the current logic trees in a different way, making the chance of no ABrk / Gnd Spoliers less likely...
NoD