i think there is some kinda problem with the opto transistor chain and i think some of the time the gate voltage of the charging mosfet is in the transition region and the transistor turned on and a large slug of current crowds through as the mosfet turns on and the output caps on the charger are suddenly discharged through these high resistance phases.
i wonder if there was some way to do a slow start on the current flowing out of the charger into the battery when the charging mosfet turns back on. i think the current is lower when the mosfet is turning off. but maybe it happens all the time and that is why they cook. but nobody seems to have this problem with 4110s. maybe put a choke on the charger output.
if you would just look at the gate voltage on the charging mosfet as the battery charges up you might see what i am talking.
you can measure the voltage drop on the optotransistor also to see which ones are different from the others. maybe it has something to do with the optotransistors or the comparators.