I plan to download and do that listening test again to check my memory and biases!
Here are a few other combinations I did not try way back when, which may be informative:
ORTF + XY in the center (my take on this is that ideally a standard near-spaced pair such as ORTF, DIN, NOS, etc, should be spaced twice as wide when a single microphone or coincident pair is used in the center. Without doubling that spacing this arrangement may or may not work as well as it could)
AB + ORTF + XY (this is somewhat similar to the primary 6-channel configuration I recommend, but likely not similar enough to work correctly for the same reasons as above but with even more potential complications. Ideally the ORTF pair should be spaced twice as wide, and the AB pair two to three times as wide as it is here).
Decca tree + XY (except for the presence of the center omni of the decca tree, this would be closer to the four channel configuration I most recommend due to the omnis being spaced wider. The presence of the center omni would probably not pose a problem except for it being positioned 30cm forward of the X/Y pair rather than being coincident with it, which may create some problems here. If truly coincident with the XY pair, the presence of the omni would simply change the XY cardioids into something more like XY subcards).
For reference, the baseline 4 microphone array I most recommend is AB omnis spaced ~1-2 meters apart with a coincident center pair (MS preferred over XY) placed about 20cm forward of the omnis. When omnis are not appropriate, substitute a directional pair for the omnis, spaced about 60cm and angled +/-45deg, or angle them PAS and adjust the spacing between them based on the resulting angle - increasing spacing for a PAS angle narrower than +/- 45 deg.