I don't think there's a good enough understanding of genetics to know how many generations. I'm guessing the estimate of a few thousand years was a complete guess. Many things impact the timing and these systems are chaotic. Even selective breeding is fairly uncertain. I don't know if geneticists could predict how many generations it would take if we selectively breed gray wolves to look like pugs.

This assumes mice would gain an evolutionary advantage by being venomous.

Mice survive as a species by reproducing fast and hiding, not by biting other animals to teach them a lesson. In order to exploit venom's potential they would need to undo their risk-averse behavior around other animals, which may not be in their best interest as a species.

The article explicitly states this assumption in the same sentence that I quoted a section from.