In the meantime I do maintain that ISATM procedure is erroneous as there is no way to read the data for AOA #3
Why wouldn't be there any way? You can read all AoA values from every MCDU as has been explained above. It's easy enough to do, and i would expect it to be detailed in the relevant procedure.
I do agree however that an AoA disagree advisory would be a nice clue about whats up.