Gutbucket, hoping I don't sound too much like the Borg here, but trying to "reconcile" a frequency response graph with a polar response graph is futile. The curves in a polar diagram are independent of the microphone's on-axis frequency response.
The purpose of a polar diagram is to let you see the similarity or note any anomalies in directional response in the different frequency ranges that the microphone picks up. For each frequency that's displayed in the polar diagram, the actual 0-degree sensitivity of the microphone at that frequency is not taken into consideration. Instead, the polar diagram shows it as "0 dB," and it becomes the reference point for the rest of the curve at that one frequency. This process is then repeated for each frequency on the list.
If they didn't do that, some of the curves might go well beyond the circle (think of a diffuse-field omni microphone whose response is up 6 dB at 8 kHz) while others would be tiny (think of a speech cardioid microphone which is down 12 dB at 50 Hz). If you wanted to make a graph like that, you certainly could, by taking the frequency response curve, looking up the 0-degree sensitivity at each of the relevant frequencies, and then applying those values as "multipliers" to the size of the polar graph for that frequency.
But in the normal way of drawing a polar diagram, the frequency response is specifically excluded as a factor because that information is already contained in the frequency response graph--so it would be redundant, and for microphones with non-flat frequency response, kind of messy looking.
As a test, see whether this way of putting things now makes sense to you: The ability to "reconcile" the two kinds of graph would only be an indication of whether the microphone has flat response or not--and the frequency response graph already gives you that information, so the polar graph doesn't.
--best regards
P.S. (added later): Looking back at your message, I see that MBHO apparently scaled each of the individual polar curves by some seemingly arbitrary factor. I agree that this is distracting, but actually in principle, it's "no more artificial" than the usual way is, where each individual curve starts at 0 dB for 0 degrees. I assume it's just their way of trying to present the different shapes without having to draw lines to label them.