Paul,
As Davey said above, I've no reason to expect linear phase from the coincident driver, but one can only hope. And if linear phase, waveform fidelity, is important, its most so in that 600-4,000kz? range. If anyone expects that linear phase response at more than one, ideal listening position, I believe one must have coincidence, at least with respect to the listening position window. And the typical passive crossover isn't linear unless a 1st order or something involving a filler driver/voicecoil. Regardless, your physical topology really does at least offer the possibliliy of linear phase over a broad frequency range spanning drivers, and over a larger listening window. I wonder what the possibilities might be of applying a DEQX (or equivalent) would be.
If you're feeling masochistic, an oscilloscope picture, miked on the listening axis, of a square wave at 1 octave above the woofer to mid crossover point. Just the mid tweet combo, the woofer could be disconnected if easier. This would give that fundamental and all odd harmonics. Low levels are fine as you don't want to fry your tweeters. (or lowpassed without phase corruption) Same thing for 15 degrees of whatever axis. I don't expect waveform fidelity, but if the signal were the same on and 15 degrees off axis, that would be really cool.
Thanks,
Paul