I've read often in the old threads that parallel setup of cells is considered safer than series.
Well, Series can have its drawbacks as if the cells are unbalanced the one that goes discharged first is in risk of being damaged from the current the other cells still can provide through the circuit, up to the point of reverse charging, which is damaging the cells and could be dangerous too, if unprotected.
But parallel is not really good for the cells too:
Any imbalance between the cells voltage, even minimal difference, will result in higher voltage cells to flow current to the lower voltage ones, trying to equalize the cells voltage. But this is not free, the internal resistance of cells always takes its toll on this current flow. Also the cells cannot really selfbalance to perfection, there will always be minimal difference, thus there will always be a low current flow between them still trying to equalize. In the long run, cells will all be depleted.
I cannot say in how much time, but know this is a sensible problem in other fields: in automotive or nautical, the multi battery setup in parallel always have electronic or mechanical (manual) separators to isolate each battery and avoid this self-equalizing risk.
What about our cells?
Is Lithium ion (any of the available variations) maybe less sensible to this than the lead accumulators used in the above mentioned fields? Don't think so.
I would like some experts thoughts on the subject.
In the meantime, I would consider a serial setup safer over a parallel, because PCB can trip and avoid problems in a serial setup, but it's totally ineffective in a parallel setup. It will not trip untill all cells are depleted. Will not really damage the cells (maybe, unsure) but one will end finding the torch totally depleted instead of fully charged. Nice when you need it, eh?