While the IV and PV curves can vary quite a bit from panel to panel, it seems like the shape of each has been constant for each panel I have seen, as Voc shifts with temperature. If that holds true, you should be able to get an upper bound for what MPPT can do with a given multi panel setup by recording the PWM performance throughout the day, and recording the Voc from each panel (only disconnecting one at a time to avoid resetting the controller) each time a PWM reading is taken.
Voc can be used to interpolate IV and PV curves from the standard curve set by temperature/illumination for each specific panel. So you can guess at the Vmpp, Impp and Pmpp that could be produced for each Voc reading taken on each panel.
Draw this curve along with the PWM curve over the course of a day to guess at what the approximate max MPPT advantage could be with a specific mix of solar panels.
MPPT won't produce at this level, but I think it is an upper bound it will not exceed.
If nobody points out a major flaw with this method, I think I will try it.
Jim