Keep high resolution!
Do you feel that's the difference? No windowing needs to be applied to impedance measurements since there's no acoustic measurement being made. Mine are routinely better than 1 Hz, as are impedance traces other builders share with me from the web. A tester would have to go out of their way to make the measurement low resolution, for no real purpose.
I understand how frustrating it is fielding questions about test outcomes from so many visitors. So, please take the following solely in the spirit of trying to help, sharing some of my experience in this area.
Vibration can create chaotic signatures in the impedance trace. For example, vibration is very hard to control in free air measurements for Thiele Small parameters, causing noisy impedance traces that look similar to the Harbeths. I was discussing this with Ken Kantor many years ago, and he shared same experience "You really need to kill any parasitic vibration in the driver to get a clean, symmetrical impedance curve"
This usually doesn't affect in-box measurements, since mounting the driver steadies the frame enough to avoid these artifacts, so I usually wouldn't suspect this. However these wiggles could be caused by a vibration somewhere in the speaker itself, for example loose driver bolts (I understand its an older model)? Or perhaps very high measurement levels causing non linearities to be excited.
As well, room reflections if large enough can also affect the impedance sweep. Please see figure 5d and the analysis here:
http://pub.dega-akustik.de/ICA2019/data/articles/001506.pdf
View attachment 47544
Comparing signatures with the Harbeth:
View attachment 47547
This shouldn't affect the substantive outcome (doesn't appear to affect frequency response measures), but just food for thought.