From my experience cell voltage under load has to drop to around 2.5V (or even lower) to trip protection circuit. After resetting protection circuit and removing load, cell voltage will be around 2.8 thus providing adequate protection to the cell. There is possibility that under (very) small load protection circuit might not work.
I kept the drop-in running tonight and finally the protection circuit tripped at 2.5V.
I'm in agreement with chiefinspectorfinch. Mine are set to trip at 2.5V, very low for my liking! It also totally screws up runtime tests. 
I know we can't trust the protection, but I'm thinking about if I hand out a single 18650 powered light to a non flashaholic that doesn't have a DMM (lol) = an over discharged cell which comes back to me, best case.
Did you measure cell voltage after resetting protection circuit ?
Voltage was 2.6V after it came out of protection.
It's good to test a cell down to it's limit like this, in a controlled way. I now know how to treat this cell.
Tripping at 2.5V is pretty terrible though, the protection circuit isn't a million miles away from being totally useless. Perhaps I should call this a semi-protected cell 
Think of a protection circuit as more of a seat belt for your cell something for emergencies (shorted cell for instance) not something you want to rely on in general use.