While waiting for gerrycan's reply I've done my own googling research on the significance of the stubs.
These are the stubs they talk about:
They are the short stubs attached to the northmost beam east of the girder between 44 and 79.
If they are that far, then what makes them so important that the whole report has to be thrown away because NIST didn't consider them for inclusion in the ANSYS FEA?
I found the explanation (by Szamboti) here:
http://aibafs.com/thread/9998/new-standard-deception-nist-report
But as we have seen, that is of no significance to the walk-off, as the movement of the column to the east introduced the displacement that, together with the expansion of the beams, would cause the center of gravity of the girder to walk off the seat, leading to failure.
Interestingly, these stubs would make that beam provide a leverage for the girder's end in the C79 side to displace more than the beams pushed it. But that's a side note of no relevance here.
So, there, that completes gerrycan's list. None of its points changes the validity of the NIST report in any way.
These are the stubs they talk about:

They are the short stubs attached to the northmost beam east of the girder between 44 and 79.
If they are that far, then what makes them so important that the whole report has to be thrown away because NIST didn't consider them for inclusion in the ANSYS FEA?
I found the explanation (by Szamboti) here:
http://aibafs.com/thread/9998/new-standard-deception-nist-report
The NIST WTC 7 report actually claims two potential failure modes to move the girder off its seats.
1. The girder is pushed off its seat at column 79 laterally from east to west by beam expansion.
2. The northmost beam buckles, after pushing against the girder end that cannot translate due to it being between the flanges of column 44, causing the other beams to buckle and rocking the girder off its seats by pulling it back to the east.
[...] In the second case, the lateral support of the beam stubs keeps the northmost beam from being slender and buckling, while also allowing it to deflect the girder to the west, while the girder is restrained by the other beams and putting them in tension, precluding any possibility of them buckling.
1. The girder is pushed off its seat at column 79 laterally from east to west by beam expansion.
2. The northmost beam buckles, after pushing against the girder end that cannot translate due to it being between the flanges of column 44, causing the other beams to buckle and rocking the girder off its seats by pulling it back to the east.
[...] In the second case, the lateral support of the beam stubs keeps the northmost beam from being slender and buckling, while also allowing it to deflect the girder to the west, while the girder is restrained by the other beams and putting them in tension, precluding any possibility of them buckling.
But as we have seen, that is of no significance to the walk-off, as the movement of the column to the east introduced the displacement that, together with the expansion of the beams, would cause the center of gravity of the girder to walk off the seat, leading to failure.
Interestingly, these stubs would make that beam provide a leverage for the girder's end in the C79 side to displace more than the beams pushed it. But that's a side note of no relevance here.
So, there, that completes gerrycan's list. None of its points changes the validity of the NIST report in any way.