Smithy looks at the timestamp and sees that the feedstock-local python 3.13 migrator has the same timestamp as the 3.14 migrator and uses the 3.14 migrator.

I thought the decision to apply is always based on the migrator file( name)s? That a randomly equivalent timestamp can change which migrations are applied sounds like a bug to me?! Or what could be a possible use-case for that?

Actually, smithy should delete that 3.13 migrator since it's not in the global pinning anymore... This is all very strange to me. 🤔

(but maybe we had a regression in this regard; the numpy2 migrator often also didn't get removed: #2336)

Originally posted by @h-vetinari in #2367