I don't know. But I do know a very tight tolerance pair matching is one aspect.
I'm not sure two matched speakers in a room will still be matched.
Even if you have two speakers matched to within 0.01dB, the typical room makes them mismatched by more than 3dB above transition, and more than 10 or 15dB below, unless you have a magical room. And even worse, the early and the reflections by large amounts as well. The room bends the speaker's response dramatically. So how can exact matching even be a criteria? I have always heard it's even smooth dispersion that allows speakers to provide a reasonable sound-field, creating the illusion of an image. I keep reading this, and I see it in good speaker's measurements.
Here are a left and right pair of well-matched speakers in a room.
These pair are within 0.5dB of each other, but are a mess in my room.
The smoothed response post room-correction isn't so bad:
You seem to have noticed the same issues in your REW measurement thread on your own speakers. Don't despair! You have great speakers, it's your room. The good news: these good speakers in a reasonably controlled room, even with the large room-induced imbalances will still energize the room in a way that allows a good image on the right recoding. I invest zero time on left/right, I do invest time on trying to tame room modes and reflections. I think I showed these effects of the room in the Perlisten thread you started. In my experience the reflected sound is even more variable, and those interactions are critical in the trick of imaging. Amir and others mentioned this earlier. There is an entire thread on imaging where this is discussed. Probably better to just read Toole and the literature.
One parting quote from Toole:
It is an interesting trait of the human ego that most of us, most of the time, know how certain things should be. Stereo is no exception. In spite if the fact that recordings are created, by and large, from multi-track tapes (or at least multiple microphones in acoustically foreign environments), many folks persist in believing that two speakers in a room are capable of rendering positions, dimensions, and ambience that, to them, should be perceived as realistic. Fortunately, humans have good imaginations, because occasionally this miserably imperfect system works amazingly well.
Yes, I am. You don't have to prove anything, just show me two separate measurements of the same exact speaker where you can overlay and the measurements are identical. That simply does not exist.
Hopefully the above L/R response graphs help show that even matched L/R pair are going to be mismatched in a room. I would be irritated if I got a pair with a large mismatch for sure. I even measure my speakers out of paranoia, but more for defect rather than L/R since I have had drivers with issues (new drivers with VC rub, a new TOTL Seas woofer with surround delaminated from the cone, etc...)