And, FWIW, I am now really done testing the FR of this speaker.
I lied.
After looking over my data more closely I realized the listening window result didn't make sense (it was higher in level than the on-axis response in the lower-midrange, which didn't make sense for this speaker). Long story short, the issue wound up being my ground-plane vertical spins. The previous setup I used implemented a wooden turntable which put the speaker about 3 inches off the ground. I wouldn't have expected this to matter because the wavelength of 3 inches is still quite long relative to the band I wanted to measure here. But I was wrong. It mattered. Just enough to fudge up the response above 300Hz - 600Hz which is smack where I was using Klippel to create the room correction curve for the far-field measurement. IOW, I was feeding Klippel bad data and it made the vertical response below 300Hz erroneous. Here's what I mean.
The picture below illustrates the difference in how the two measurements were taken. The first is the old vertical measurement; speaker off the ground. The second is the new measurement with the speaker laying on a very thin piece of pegboard about 1/8" off the ground.
Now, here is the response difference taken at -10° vertical. These are GP only. So don’t consider them for full range response. Green is the new measurement. Red is the old measurement (taken on the turntable, a couple inches off the ground). Both were tilted toward the microphone for proper HF response (to some degree; I only care about <1kHz for GP measurements).
Amazing how sensitive these measurements are to assumedly minor variations even in low-to-mid frequencies. I've not had this issue with other speakers because they didn't require building up a platform to easily spin around on the ground, which I did for this speaker. The more I measure, the more I learn and - I have to be honest here - the more I wonder just how capable others are of doing these kind of measurements. I realize that sounds egotistical but there is a reason. This stuff is not trivial. You have to really understand the measurement setup, the drawbacks, the quirks and take the time to analyze the data before just tossing it out there. So, if that does come off as egotistical, so be it.
Anyway, after I caught this issue, I set up everything again and re-measured all my vertical ground plane measurements. I have updated my data which now looks very much like Harman's spin data other than the response <100Hz which I believe may also be factor of SNR (measuring outdoors vs measuring in an anechoic chamber).
This thing has kicked my butt. But I tell you one thing... this data is effing bulletproof now. It would hold up in court. LOL