carlito, "where the microphones are pointing" is crucial when you're using them as spot microphones--picking up a single instrument or voice at close range. But when you're using a pair of microphones as the main (or only) microphones for stereo recording, the area of overlap between their pickup patterns is what you have to aim, three-dimensionally, at the totality of direct sound sources in the situation.
That's much more complex, and "where the microphones are pointing" is certainly one of the variables that you have to adjust. But the microphones have to work together, not independently, to create a stereo image. As a result, they aren't individual "target-seeking" devices; they're more like parts of a larger array of antennas, where the point is to keep the entire array open to a certain range of pickup angles where you hope to find signs of life in the heavens.
Anyway, no microphone is so narrowly directional that it picks up sound only on its own axis. Please forgive the metaphor, but if a microphone really was a penis, even the narrowest shotgun microphone would still require a lot of cleaning up all around it after every use! Microphones are much more analogous to receptive sensory organs of some kind (not that a penis isn't one of those, too). A cardioid pattern is practically almost an omni except for behind it; anyway its directivity isn't nearly as strong as what most people seem to imagine. And now I think I will leave that metaphor behind, thank you very much.
Generally in stereophonic playback, we don't want any direct sound sources to come at the listener from the exact position of either the left or the right loudspeaker. That somehow destroys the effect. The effect works best if the stereo "sound stage" seems to extend all the way between the loudspeakers without ever being in either of them as such. And that means that all the direct sound sources have to reach both microphones at angles where they are still fairly sensitive, i.e. using either omnidirectional microphones which are sensitive in all directions to begin with (and not too far apart), or else directional microphones that are positioned and angled so that the whole set of direct sound sources falls within their common range of sensitivity.
Does that make some sense?
--best regards