Intrigued by the extra connection in between the positive and negative, and suspecting of some kind of temperature dependent indicator (before I seriously googled about it...), I decided to dismember one of the batteries and check if this was attached to some temperature dependent resistor.
To my dismay, after carefully removing the top of the battery, I found a small PCB inside:
(Sorry for the barrel effect, but the photo was taken through a jeweler lens. This is a tiny PCB, measuring 20x5 mm)This is by no means a NTC or PTC resistor... This is a well crafted protection circuit, as well as a pair of resistors that inform the outside world of the type of battery this is.
The thermal protection I expected to find, is not part of this circuit, rather takes the form of a thermal fuse, that goes off at some predefined temperature that I could not determine, as I tested its efficiency by using brute force methods, meaning "put soldering iron tip close to the fuse till it conducts no more".
So, driven by an increasing curiosity, I decided to find out what were these ICs and what role they played in the protection of the battery.
Armed with Google and a lot of patience, I found that the leftmost IC, marked 2450B, contains a pair of MOSFET transistors, while the center IC, marked M5A581, is designed with a few goals in mind, namely: avoiding overcharge, short circuit and overdischarge situations, that might endanger the cell.
I managed to find the datasheets for both ICs, from NEC and Seiko. The ICs on my battery were certainly manufactured by some obscure factory somewhere in China.
NEC 2450B Datasheet
Seiko s8201 Datasheet
The circuit depicted on page 25 of the Seiko datasheet is the one implemented in the battery. I don't have the means to measure the capacitors, but I guess, by their colors, that they match the schematics on the datasheet.
As for the mechanism used for the BSI (Battery Size Indicator) pin, I can only guess it is a resistor placed between this pin and the negative pole of the battery. My eyesight and my rude lenses don't allow me to fully check the circuit.
I'm on my way to design a simple and effective circuit to charge these babies, knowing beforehand that they are a bit more forgiving than the noname nobrand cells we usually find inside our gadgets.