Axelwahl your last point is where things get interesting.
Like everything else it is a matter of compromise. As you have noted, a low inertia arm reduces the maximal VTF variation. The compliance required to keep the resonant frequency in the right range changes at the same time, so the effect of a given warp in terms of displacement of the cantilever suspension depends on the resonant frequency: the higher the resonant frequency the smaller the effect.
Unfortunately we're not free to move here. As previously noted the product of inertia and (rotational) compliance forms a low pass filter. As the equation previously given shows, the attenuation and phase response of this filter depends on the ratio of f/f0, so as the resonant frequency moves towards the audio band the effects of these become more and more pronounced. That is why resonant frequency is optimised over such a narrow range.
Mark Kelly