Bulgie's argument is that the center of the stay will not see as much elongation or compression, which is pretty much the entire the basis of using tubular structures, thus a sharp crease there won't see the stress/strain as a sharp crease near edge would. Note that this doesn't apply to compressive loads along the drive side chainstay, only the bending and torsional loads. Thus a sharp crease near the center will see ~0% strain under torsion and bending, and a sharp crease near the circumference will see some arbitrary value X% strain under torsion and bending.

That being said, I don't see any particular reason to chance it. The wall thickness on Barras are a mystery. Letting a tube crack then welding it is never following best practices. The crease gets closer to the circumference near the end of the flute. If I did do this on 853 and it didn't crack, I'd torch it and harden it again in the dented state.

Also, I've thought about the heat-treatment process some more. 631 tubes might not be annealed, but stress relieved (essentially a high/soft temper) after drawing like other non-air-hardening steels are, which would mean getting the temperature high but not past critical could offer the ductility of 631.