So the overall combined sensitivity pattern of a pair of X/Y cardioids is always more or less semi-cardioid in shape, specifically to an extent determined by the angle between the two microphones.
If you place that X/Y pair of cardioids in the center between spaced omnis, and pan both channels of the X/Y pair to center, the practical result is the same as if you had used only three microphones and placed a single semi-cardioid pattern microphone facing directly forward between the two omnis.
If you pan the X/Y pair outwards away from center, the shape of the overall combined sensitivity pattern of the X/Y pair won’t change, but level differences between the Left and Right channels are introduced. Because the two X/Y pair microphones are mounted in the same physical location in space, there is no phase difference between them, so the phase relationship doesn’t change regardless of how they are panned. That’s not the case for a near-spaced pair.
Because the phase relationship between the two X/Y channels never changes, regardless of their panning, the phase relationship between the X/Y pair and the omnis doesn’t change either. It only changes if you physically change the spacing between the microphones.. or if you pan the omnis in towards center but don’t do that. Just leave them routed hard-left and hard-right. The X/Y pair will fill-out the center nicely.
If you use a near-coincident pattern in the center instead of X/Y, the phase relationships are considerably more complicated even if you don’t try to pan anything around and simply mix the left cardioid directly with the left omni and the right cardioid with the right omni, both hard-panned left and right. If you want to adjust the panning of either the central near-spaced cardioid pair or the omnis (not sure why you’d want to pan the omnis), then you’ll introduce comb filtering artifacts similar to what you get when attempting to mix a near-spaced pair down to mono.
The mono-compatibility of X/Y (or any other coincident configuration) is what makes it a very good choice to use between spaced omnis, not because the end result we want is a mono compatible mix (it won’t be, because of the spaced omnis), but because the same problems encountered when mixing stereo microphone configurations down to mono are encountered in a multi-channel scenario when mixing more than two physical microphone locations down to two-channels.
With three microphone locations it’s manageable and not that problematic mixing to two channel stereo; with four locations the phase relationships become quite complicated. In addition to being more potentially problematic to begin with, that limits the degree of freedom you have when mixing it.