I think William is correct. Now that's not absolutely certain, but if that is in fact a phase induced notch, you should be able to make it go away. (Kudos on being able to actually have a phase notch show up in your measurements, as it's most often obscured by modal response with 'in-room' measurements from the listening position. The close proximity of your subs to mains, and having a couple of subs for modal smoothing may account for being able to actually measure it. You're in the home stretch with this.)
That sort of notch will appear if the hp and lp are slope and level matched but 180 degrees out of phase. If the subs have a simple two position phase switch, take a measurement with it toggled to the other setting first, before changing sub distance in the AVR. If that does not reduce or eliminate the notch, then do as William suggested and add distance incrementally and re-assess until you get there.
If you can make the notch go away by adjusting the phase/distance settings, next look at impulse response. You'll find that you will get flat freq response, but the impulse response will likely be off, a full cycle or an integer multiple of that at the crossover frequency, which is more than a simple polarity switch on a sub can accommodate. You can then use the AVR distance settings, which can accommodate this, to restore or achieve optimal impulse response; it may help you dial in impulse response to add distance to the mains rather than to the subs depending on which, hp or lp is leading. The distances you set in the AVR may have no relation to actual distances in your room, or they may. Trust the measurements when doing this.
You get all that taken care of and you'll start to notice more resolution and detail in the critical mid-bass band than you realized was possible.
Excellent progress, Pogre!
