Your experimental results are very confusing. I can't make head nor tails of it. But some observations:
Voltage is a very imperfect, maybe useless in some respects indicator of state of charge. Apparently your solar panels kicked up the voltage from "full" to 13.4V from 3:00 to 4:00. Then 45 minutes later it dropped to 12.6. What conclusions can you draw from this. Almost nothing as the solar panels were affecting the battery voltage.
Note that when you turned the fridge back on in the morning that the voltage immediately dropped from 12.2 to 12.1. Did the state of charge drop then? No, the fridge load caused the battery voltage to drop.
My point is if you are going to use battery voltage as a broad measure of state of charge, you have to let the batteries rest for at least an hour with no load or charging. Otherwise the load or charge effect totally distorts the voltage and makes it unusable as an indication of state of charge.
David
__________________
|