Sully's views on a third AoA should not affect the return to service - cf EASA position.
If the effects AoA error/failure are fully contained by the modifications, then an additional AoA may not improve safety beyond that of the NG. i.e. if Max requires three AoAs then why not previous variants.
The concerns about the alerting could be more persuasive, not just the indications, but the underlying logic and operation of the modified systems. e.g. the Speed Trim Fail annunciation appears to have two meanings; it could alert that MCAS has failed (and / or STS?), or as information that MCAS is using single source AoA and is still available.
The descriptions of the modifications given so far (FAA Summary) do not suggest that issues of workload and confusion resulting from multiple alerts (e.g. stick-shake) due to AoA failure have been improved. This may also be put aside as it is not an MCAS issue in isolation; as the Max / NG debate above.
There could be additional confusion from the changed drills and training; the Speed Disagree alert has no drill and refers to the Unreliable Airspeed Drill implying that speed is unreliable - where the definition in that drill is subjective - what is unreliable or not.
'Disagree' relates to DADC/ EFIS comparison; if this is due to AoA input, then is the low speed awareness similarly in affected, or would the new AoA logic change that - but to what.
See training thread; how can training be effective without information.
Is the FAAs reluctance to disclose background reasoning for approving the changes restricting the level of system knowledge which crews can achieve.