So in terms of further general testing I put the modular through my Yamaha AX-497 Stereo Receiver and directly out to headphones from there ... I don't know if this is "professional" level equipment, but it's the only other real output I have for the mix besides the MG10Xu and my impression is it's a generally HQ receiver. Certainly cost way more than the MG10xu anyway! But anyway this in and of itself made a difference in terms of how easily noticeable the sound was, it was basically inaudible until about the 3 o'clock point on the rotary volume knob, whereas audio levels generally are about as loud as you can comfortably listen to by the 12 o'clock mark. However, the sound was still there, and as I was trying some patches and such, it seemed like it became an interference after a bit, like suddenly there was this faint background hum to the patch that wasn't there in the beginning just out of the blue.
So I began changing some modules in and out, double-checking my connections and whatnot, and eventually found a configuration in which the sawtooth-like noise was not there, and at the 9 0'clock and above level all I would get was a bit of white-noise/hiss from the receiver itself (which I'd expect with the volume at that level). Then I started experimentally adding modules back in ... and eventually would hit a point where the sawtooth-like noise came back. I switched a few things out wondering if it was possible a particular module could be causing it, and if so trying to pinpoint which -- but it seemed to vary and right now my impression is that it's once the power draw is past a certain threshold, that the sound becomes present.
I have quite a few modules, but I have an excel sheet calculating the power draw of everything against the specs of the PSU, and in theory none of the rails should be at more than 50% of what that PSU can supply power for even with every single module installed, soo ... in a way I'm back to square one, but until I learn a bit more about how to check or correct the PSU itself, at least now I have a way to mitigate the issue temporarily (which is basically to only use about half of my case's available space).