Dug into the pattern charts a bit. I see that MS with two figure 8 gives you the ability to virtually change the angle of a pair of Blumlein figure 8, after the fact, with a change in the blend. Another win. When you change Blumlein width with panning you only change the presented width between playback speakers, as opposed to a virtual pattern repositioning. This would seem to change the expected relationship of the combined pattern nulls, and explains the changes in the high frequencies. It ideally shouldn't be panned anywhere between outputs at all. All theories with ideal poor patterns across all frequencies, of course. The real world will treat us differently.
Yes, Mid/Side using two figure-8 pattern microphones is unique in that the shape of the resulting virtual polar patterns doesn't vary along with a change in Mid/Side ratio, only the virtual angle between patterns varies.
We're getting quite off-topic discussing Mid/Side here though. This thread is specifically about a
spaced Point-At-Stacks microphone arrangement using bi-directional (fig-8) pattern microphones, not about about Mid/Side. I'd prefer not to confuse those who may not be familiar with both techniques. Plenty of threads about Mid/Side here at TS, lets keep this one focused on PAS figure-8's, Thanks!