Calvin

On these Pylons SOH is a very crude, hard-coded estimate.
For the first 1000 cycles: for each 100 completed cycles the SOH is decreased by 1%.
Thereafter, for each 167 completed cycles the SOH is decreased by 1%.
SOC is calculated based on Coulomb counting (from full) using the current SOH.
The exception to this is when charging above 89%, or discharging below 11%, SOC is based on a lookup table using the lowest cell voltage. That is why, when charging, the SOC seems stuck at 89% and then suddenly shoots up.
A further exception is that, when discharging, if any cell voltage gets into the values contained in the lookup table, the SOC reported for the pack is that value. This is why SOC will suddenly drop very quickly if there is a weak cell in the pack.
I have the full set of voltages and corresponding SOC values available if anybody is interested.

Below what you get from a 30° panel.
 

And a 5° panel...

As expected the 5° panel does slightly better in summer and not so well in winter.
The shading in your current setup affects not only the panels in the shade: all panels in a string where 1 panel is shaded are affected.  If the strings are all on one MPPT the losses are even greater.
So I would agreed with @YellowTapemeasure - flatten the front row.  The losses are probably small compared to the shading losses.