So, first, this disclaimer: I don’t have an Iridium (yet), but I’ve been hungrily following this thread and appreciated your contributions.
One thing it could be is, as you mention at the end, a sort of “sample rate mismatch.” The Iridium may only sample its CV inputs at a certain rate that is somewhere below audio rate (if the designers did not expect audio rate modulation, this would be a way to save CPU – you can sample a sine wave moving at LFO speeds well below audio rate and still have much higher resolution than MIDI, for instance, so no zippering, etc. artifacts).
The cyclic aliasing could be the result of mismatch in rates producing phase shift in the sine wave, as different samples would begin at different points in its cycle (and maybe the strongest evidence toward that is that we hear the aliasing rise and fall like a peak of a wave). Kind of like those old flip books which would mimic animation – a horse galloping and so on?
Just a theory, though.