When the situation allows for it, the microphone sensitivity pattern I most prefer in a 2 channel near-spaced stereo pair leans toward the more open, omni side of cardioid, of which Schoeps MK21 and MK22 are excellent examples. There are a number of reasons for that, several which are generally recognized around TS and have to do with the attributes of the individual microphones as well as the room and audience situation where the recording is being made. Specifically regarding pickup pattern behavior, I cant remember where, but recall reading that the MK22 was specifically designed to very carefully manage the transition from primary control of pattern via differential-pressure over to control via acoustic pressure boost on-axis at high-frequencies (a result of the geometry of the diaphragm face and capsule housing) in such a way that a narrowing of the pattern at the highest frequencies typical of other microphones is minimized, resulting in a pattern which remains more consistent than it would otherwise. DSatz, Is that accurate?
A few of the following statements may ruffle some feathers, but I stand by them. Feel free to disregard.
Another reason, which I think significant even though it isn't discussed as much around here is the combined sensitivity pattern of both microphones together in a stereo pair. I've found I like a collective sensitivity bias that's generally subcardioid shaped, and oriented toward the performers of course. How open depends on the situation, and good pattern behavior in the individual microphones generally makes for better collective pattern behavior. Perhaps the only thing in that statement that may seem odd is the talk of the collective sensitivity pattern of both microphones working together.
Here's the part that may spark some controversy- For the most part I achieve an overall sensitivity pattern that trends somewhere between open-cardioid/sub-cardioid-like by combining a number of supercardioid microphone channels that are oriented in different directions. In addition to a pair oriented +/-45, there is a third facing directly forward, and an additional pair facing rearward. When mixed they collectively produce an overall sensitivity pattern grossly similar to that of a pair of subcardioids in a typical near-spaced pair arrangement. I think that is significant and I hear a lot of similarities between the two, and that's really the point of this post.
Of course doing it that way is more complex, and of course there are pitfalls. It's not that I'm proposing my method for use here so much as trying to convey an interesting observation about a fundamental aspect of live performance recording using a microphone configuration placed on stage or out in the room. I went the more complicated route for a number of reasons, one of which is that it enables alteration of that collective stereo sensitivity pattern afterward, and generally provides other forms of flexibility which help me to more consistently achieve the kind of sound I'm happy with.
The more typical way of exercising some of that flexibility is by substituting microphone capsules of differing directivities. Swap MK22 for MK4 or MK41.. or run multiple setups simultaneously. I achieve overall sensitivity pattern through the manipulation of multiple microphone channels. I'm unable to afford four or five MK41 in substitute for a pair of MK22, and of course in many other ways that substitution is not equivalent, but that is the general idea behind the observation.