If you use a pair of back-to-back, separate cardioid microphones to simulate a figure-8 pattern, their diaphragms must be at exactly the same distance from the sound source(s) or else there will be phase conflicts and cancellations at high frequencies. Even a small amount of distance between the front and back capsules can make for terrible off-axis high-frequency response.
An example of this would be the "vintage" Neumann KM 86 (see attached photo), which had a pair of matched cardioid capsules (the same ones used in the KM 84, and later in the KM 140 and KM 184) back to back inside its capsule head. But they were separated by about half an inch internally, and the resulting response at 45° was peaky and shrill. Notice how in the attached polar diagram at 45° there's 4 dB of increase in the response at high frequencies which are already up 2 dB in the on-axis frequency response graph, for a total peak in high-frequency response of ca. 6 dB. In a Blumlein setup, the center of the sound source is being picked up by both microphones at 45°, so the evenness of response at that angle is crucial to the result. (And yes, I learned that the hard way, by trying to record an operatic soprano in a small, bright room with a Blumlein pair of KM 86s.)
The problem of coincident placement is easy to solve if you only need one figure-8 microphone (e.g. for M/S stereo recording), since you can place the cardioids "head to head" if they're side-addressed or overlap them vertically if they're front-addressed. But if you're trying to create a Blumlein pair, I don't see how it's possible to fit four side-addressed cardioids together closely without specially-wired dual-diaphragm capsules (e.g. Neumann QM 69, AKG C 424), which compromises the imaging at low frequencies since dual-diaphragm cardioids always become wide cardioids in the bass.
If your cardioids are front-addressed you can use a kind of "X" arrangement with all four microphones pointing in from the corners of a square.
--best regards