In thinking about all this, it occurs to me that imaging conflicts between stereo pair spacing (time cues) and angling (level cues) will be most significant with regards to near-spaced microphone configurations, and won't negatively affect coincident nor wide-spaced AB configuration imaging to the same degree.
With respect to coincident arrangements, even though one should strive for the maximum practical coincidence and preferably err toward mics pointing away from each other rather than across each other when not vertically arranged (one placed directly above the other), lack of perfect coincidence is more likely to affect things tonally (via combing determined by the non-coincident distance between microphones) and in a general sense of producing less sharp imaging rather than causing an outright conflict between perceptual imaging cues. Maximum time of arrival difference between channels of a close but not perfectly coincident arrangement can be significant enough to produce combing at higher frequencies, while not being large enough to produce a significant perceptual image shift.
It also won't matter much for very (overly?) widely spaced arrangements, where angling the microphones inwards can actually help unify central pickup and solidify the middle by placing center sources more on-axis with both microphones and by reducing the level difference between channels for off-center sources to some degree.
For the most part, in both the coincident and wide-spaced cases perceptual imaging is dominated by one imaging cue over the other.
Its near-spaced configurations where two relevant aspects come into play:
1) Near spaced configs are intended to achieve a certain degree of balance between timing cues and level cues- both are intended to work together. This makes direct conflict between the two types of cues are problematic, at least with regard to imaging.
2) And now I can't remember the other one.. uh, time to go eat.