For your information Charles, I modified the circuit to charge only one 4.2V Li-ion battery (see attachment) and I measured following conditions when the leakage happens -
1. MPPSET voltage is set to 6.1V while VIN is given 5V
2. Both high and low side FET are off, VGS < 0.1V
3. All loads to other circuit are completely cut off.
4. Total current draw from battery is measured 6.4mA, the voltage across 0.047R sense resistor is 15mV, so current thru it is 3.2mA.
5. Although i can not measure current thru the inductor and FETs, it can be deferred there's no current. Because (a), VIN(5V) is higher than VBAT(4.2V), and (b) FETs are all off, there can't be current thru it.
So i conclude current is going thru both SRN and SRP pin, about half each.