The batteries will be charged by the charger with the higher voltage potential at the connection to the circuit, respectively. Even though the GP was showing 14.7V, that isn't necessarily the voltage at the connection to the circuit. Since the GP stepped back, the WFCO provided the higher voltage to the circuit (14.4 for a standard WFCO) and dominated the charge. The GP saw this reduction in electron flow and switched down to float mode (13.5 V).
Now that the WFCO has an even higher differential to the GP and if a circuit is complete, electrons will flow backwards to the GP. It seems that GP has expected this reverse polarity situation and rather than providing just a diode which can burn out if the back flow is too great, they have also provided a resistive circuit to convert those 2 amps of electrons to heat thus providing reverse polarity protection.
The random GP manual I looked at state the maximum wire size is 14 AWG. Depending on the amperage the panels were charging at, it wouldn't take very many feet of wire between the controller and the charging circuit to loose 0.4VDC and assuming no losses in the connection itself.
5A - 0.4V lost in 16 feet of wire
10A - 0.4V lost in 8 feet of wire
20A - 0.4V lost in 4 feet of wire
etc.
