I seriously think that NIST's model had as its initiation mechanism the unseating of A2001 from column 79. This was a singular event. If this initiation was impossible, the rest of NIST's model fails.
NIST describe it as a probable sequence. Even if Hulsey's study shows this connection was not the first to fail, then that does not mean the building could not collapse from fire. Lots of damaged connections were identified. And if I remember correctly the macro simulations don't even show that girder falling first.