I guess there’s also a threshold of “too much of a good thing”. Iron oxide is significantly less permeable than iron. If the rust penetrates deeper than just the surface of the laminations, it would lower the permeability of the core.
But it’s an interesting idea. First I’ve heard about surface rust reducing eddy currents. From experience, I’ve used a quite a few rusty transfoermers that sound great. Especially old Jensen JE-123-S in API circuits. Seems like just about every one of those Jensens from the 70s have a healthy layer of oxide on them.