...
It looks like the solutions are probably fine for most, if not all the antennas (except possibly HexE23, since there's a lot of power in the off-diagonal terms, Dpq and Dqp), but the curviness of the phases makes it hard to tell. Therefore, I will choose a reference antenna to plot all the phases relative to that. Since there's a bunch of tiles whose phases are strongly sloped (in near the middle of the plot), I'll choose a reference tile with a negative slope as well, but whose gradient is not as steep... HexE25 will do. To do this, I need to re-run the plotting script with the following change:
...