Thanks to Rudolf Finke, I’ve realized that I can do off-axis simulations with the Edge, by changing the microphone distance and horizontal offset. This is important to ensure that the polar response of the drivers is not mismatched at the crossover point. Here is the simulated response of the woofer at angles of 0, 15, 30, 45, and 60 degrees:

Note: the microphone is at 3m distance and 1m height. It’s not on the woofer axis.

Here is the simulated response of the midrange at the same angles:

From these curves, it look like a perfect crossover point to match the dispersion of the two drivers would be about 400 Hz. To push it up, a narrower baffle would be needed.

It does, however,  look as though the crossover from the mid to the tweeter at 5 kHz is going to have a significant change in dispersion at the crossover point, if the Neo3 is used. Perhaps another argument in favor of a compression driver and horn.