My 2 cents:
Dayton UMM-6 and EMM-6 look more alike than different, except that the UMM-6 is, for lack of a better description, one noisy sucker. What was its input gain set at? They must already be using a FET in source follower configuration (think Linkwitz mod) in order to meet the published top level of 135 dB SPL at 1%.
The SM58 has essentially no inherent distortion (what's there might actually be from the mic preamp - what was it?), but obviously its response really tanks up top.
The 7052PH obviously delivers the best results, but at the price it better do!
(BTW, I think I have a mic that's arguably worse than all of them, a Monacor ECM-40. Self-noise in the 40s dB SPL, plenty of second harmonic, and that as an analog mic for XLR. I think it's a really old construction, it's got this 1980s feel to it. My current Sonarworks mic didn't come with as much cal data, but is so much less noisy it's not even funny.)
I might have tossed in a random (cardioid) LDC for fun. From what I can tell, even inexpensive ones tend to have good IMD performance (using a clone of what is apparently a classic Schoeps circuit), while SDCs with their more restricted internal space tend to fail a "key jangling" test more easily - I suppose they are more often than not using a single-ended 2-transistor affair, à la Oktava MK-012. I imagine the Rode NT5 would already be more refined.
It is interesting how you basically can't see any IMD from the tweeter itself. I wouldn't have expected the microphones to give up first. Mind you, there's always the option of increasing the distance, self-noise permitting, but...