It's clearly a mysterious phenomenon... but I think it may be a whole combination of things.
This may be one of those cases where shining the darkness absorber on it could provide some insight; sometimes knowing what something is, is more about knowing what it's not...
I think we agree that this does not occur in an air core configuration, at least not to an appreciable degree. That would indicate that the core plays a role.
So what are the differences between an air core and an iron core system?
Aside from the material that the core is made of, there's another aspect that I hadn't
completely thought about when pondering this... the air gap.
I went looking around and found
a PDF that is about a motor. While the document is not specifically geared at generating power, I think we can agree that the principles are very similar, if not identical, just in reverse.
In the PDF, he mentions (on page 9) the concept of the air gap storing 'excessive' energy. Applying the concepts mentioned in the PDF, in an air core system, the coil is
part of the gap, and in the iron core system, the coil is outside of it. At the higher frequencies where the phenomena takes place, this could easily be explained by the lowered permeability. It doesn't show up at lower frequencies/rotor speed because the permeability is higher, and the field is reaching the coil more effectively.
I think also some of the confusion is coming from the term 'frequency'. You mentioned "did not change F at all"; and you wouldn't need to. In the case of the mentioned mill, running normally as it would usually be and then being shorted, the frequency would only change if the rotor changed speed. It's not the change of frequency that makes the difference, it's the sharper transitions that the core experiences
due to higher frequencies.
This leads me to think that the scenario arises from interaction between the coil and the core, the core and the air gap, and the core and the magnet
across the air gap, with hysteresis being influential at some form of boundary where the coil and core 'meet'.
Hysteresis gets involved because possibly a heavily loaded/shorted coil may be enough at a high enough frequency to prevent the field from propagating through the core altogether, thanks to the back MMF provided by the shorted coil. The energy would then just be spooled up in the air gap between the core and magnet, with nowhere to go, and is returned (like a spring) to the rotor as it passes, minus losses.
Almost like reverse cogging. Instead of attraction, it's repulsion.
Steve