So by flipping the polarity on the center channel we know that this cures the dip in the near field, as shown in that measurement above, but have you measured the center channel with the polarity reversed at the main listening position to see if that cured it?
I see where you are sitting 14' back from the center channel. Do you have a coffee table or any other furniture in between these two spots? You might look into some OC703 panels on the side walls. The best thing to do would be add some on the ceiling as well, but with a vaulted ceiling, and the WAF, that could prove hard to do!
I am still interested in finding out why the measured response is so different from the modeled response? What modeling program did you use?,