I know its been discussed before - and was one theory proposed by BBC/Nova - but would the speed change from M0.82 to M0.80 have been complete before the UAS? If pitot drains simultaneously block first the IAS over-reads, then if total pressure ram port blocked speed is locked in - UAS may only be detected when drain holes asymmetrically unfreeze. If this happens while in transition true air speed could have undershot target - and A/P pitch up to hold altitude? So out of trim when A/P A/T disconnect requiring immediiate manual correction?