I simply turned each control to the right to 3:00 o'clock position and showed the frequency response. Since you are not likely to want to boost them even more, I think that is a reasonable demonstration of what it does. At the risk of stating the obvious, I have no control over what the box does. The fact that it reduces the two ends for mid-controls is a confusing design element in the unit itself, not a problem with the graph.
I think your software has generates a misleading graph. It's showing the difference in frequency response between all knobs centered and any given knob turned, yes. The problem is that it's representing all the results as relative to 1khz.
That is, if the DUT boosts 1khz by 5 dB and doesn't change 100hz or 1khz, then the graph seems misleading if it shows that 1khz didn't change while 100hz and 10khz dropped by 5dB. It's not really
wrong (in a relative sense), just very hard to parse.
These numbers chosen only for illustration.
The third knob on the Loki Mini will either boost or cut with a center frequency of ~2khz, and has very nearly no effect on the output of 20hz or 20khz.
Your graph makes it appear that the third knob has nearly no effect on 2khz and in fact drops or boosts the extremes of the frequency band. That's not how my Loki Mini or my Lokius work.
My graph is on the first page here, so it should clarify for most people, I hope.
Anyways, the testing work is much appreciated.