Along the lines of what Gums is talking about, It should be possible to measure roll rates achieved for small control deflections and then properly set roll gain after a loss of airspeeds thereby avoiding having to go to a roll direct law (Alt 2). Subsequent control inputs would update the gains. If in dead still air, the system could periodically recalibrate with small doublet inputs.
If wing flexibility causes delays in sensing, you could put accelerometers in the wings to get more timely information.