A few cautions in what we are seeing in John's measurements. You notice that he states that these are his best equalized measurements to date. Likely the EQ accounts for lifting of any bass rolloff, flattening room peaks and dips, raising high end falloff of the response, etc. One can argue that with enough EQ you can make any measurement look near perfect as John has done. What we really need to see are the unequalized performance of the array to understand what is happening and how it really performs. John needs to explain the EQ that he added as well to create near perfect plots. Furthermore, looking at the unequalized raw data will yield a sense of the phase changes between data points.
His data reminds me of what happens when I use my DEQX system. Essentially you have near perfect response with little amplitude, phase, or time errors with that system.
What matters is the sound reproduction that is created with John's arrays and does the EQ'ing create issues of its own because of compression and off-axis phasing?