Typically we don't consider epsilon naught to change as we move around in a gravitational field, we consider spacetime itself to change and leave the Maxwell equations unaltered. So that doesn't explain away the current in a wire or rotating shaft versions.
Really, the resolution of all of these problems is left to general relativity. Redshift is a consequence of gravitational time dilation, or the fact that if you bring a clock high up in the air and then down again, it will tick off more time than a clock you leave on the ground. If we consider each tick to be a spent pulse of energy, then it's pretty easy to see that the clock that went up and came back down spends more energy than the one on the ground.
Physically speaking, this would mean that if the shaft was, indeed, rigid, then it would actually be spinning at a lower rate if you went to the top and timed it with a stopwatch than it would be if you timed it at the bottom (your stopwatch is running "faster" higher up in a gravitational field, relative to the bottom). Similarly, the rate of electron arrival would be measured to be lower at the top than the bottom, etc.
NB: these rates I'm talking about are the locally measured rates, i.e. hz as measured at the spot with a properly functioning stopwatch. The shaft doesn't need to twist or anything for these rates to differ, in fact, our assumption is specifically that it does not twist.
So when the frequency-based energy is obtained at the top and converted to mass-based energy (which doesn't undergo redshift, but instead exchanges kinetic and potential energies), the "exchange rate" between the two is based on the local measurements of frequency, and we end up with less mass created at the top than we put in at the bottom.
Problem solved, as long as you trust the claim that any such energy transmission that doesn't suffer the kinetic energy penalty that normal projectiles suffer in a gravitational field has to be frequency based and suffer the time dilation effect. Briefly, energy is the time component of a four vector, so it has to be affected like this, no matter how you try to transmit it, but I'll leave that for another rant...
That's a good point -- to avoid confusing people, I should mention that my explanations above are sort of the mirror image of how most physicists would think about this situation. I was trying to be careful to say "appears to be" instead of "is" to leave this interpretive door open :)
As you say, the most common interpretation is that time itself slows down as you enter a gravity well, leaving all the physical constants the same (since their locally measured values will appear the same to you).
But if you look at things from an imaginary "absolute" viewpoint outside of spacetime, you could think of spacetime itself as a sort of fluid that's denser in regions of high gravity. Light travels more slowly in this denser fluid, and since epsilon zero is inversely proportional to the square of the speed of light, you could interpret it as being larger where gravity is higher.
I guess you can tell that fluid dynamics was always easier for me than differential geometry -- whenever I see a system of nonlinear partial differential equations (like general relativity), it's just easier for me to think of them this way.
Really, the resolution of all of these problems is left to general relativity. Redshift is a consequence of gravitational time dilation, or the fact that if you bring a clock high up in the air and then down again, it will tick off more time than a clock you leave on the ground. If we consider each tick to be a spent pulse of energy, then it's pretty easy to see that the clock that went up and came back down spends more energy than the one on the ground.
Physically speaking, this would mean that if the shaft was, indeed, rigid, then it would actually be spinning at a lower rate if you went to the top and timed it with a stopwatch than it would be if you timed it at the bottom (your stopwatch is running "faster" higher up in a gravitational field, relative to the bottom). Similarly, the rate of electron arrival would be measured to be lower at the top than the bottom, etc.
NB: these rates I'm talking about are the locally measured rates, i.e. hz as measured at the spot with a properly functioning stopwatch. The shaft doesn't need to twist or anything for these rates to differ, in fact, our assumption is specifically that it does not twist.
So when the frequency-based energy is obtained at the top and converted to mass-based energy (which doesn't undergo redshift, but instead exchanges kinetic and potential energies), the "exchange rate" between the two is based on the local measurements of frequency, and we end up with less mass created at the top than we put in at the bottom.
Problem solved, as long as you trust the claim that any such energy transmission that doesn't suffer the kinetic energy penalty that normal projectiles suffer in a gravitational field has to be frequency based and suffer the time dilation effect. Briefly, energy is the time component of a four vector, so it has to be affected like this, no matter how you try to transmit it, but I'll leave that for another rant...