Just as one can forget to change over non rechargeable cells. I really see no benifit in this regard.
I can't say I agree with that. Let me put it in this way: I come to a dive site and I have an armory of batteries and accumulators for many gadgets like camera, another camera, main light, backup light, etc. How could I be sure which accumulators are recharged and I didn't forget anything? So, before series of critical dives I just open new disposable battery package and load the backup light. Even in hurry - which is the main evil impacting safety - I'm quite safe if I did just two things: loaded new batteries, and checked the light switching it on/off. With accumulators the third thing to be added - charging. But this thing couldn't be made quickly and seamlessly with first two..
There is one more argument against particularly Li-Ion - let me mention it a bit further.
Finally, I'm not technical cave diver, but I respect their experience and try to learn from their approaches to safety. Most of those I know through forums use disposable batteries in their backup lights.
I have been using Li Ion cells for DX for a while now. I have not had one fail that wouldnt have failed if it was a better brand.
Glad to know that you had no problems with them. But I know people with the opposite experience. Anyway, neither mine statistics nor yours allows making common conclusions here. My doubts are based on experience with electronic circuits from DX - very often the quality of assembly and/or components is awful and I would never entrust my life to it.
One more argument against Li-Ion per se is the instability of charge depending on ambient temperature. The problems will come if air temperature is below zero (Celsius) - Li-Ion accumulator must not be charged in such conditions and should not be expected for long run time. So, I'll never take a Li-Ion accu based backup light to an ice-dive.
No driver? Thats interesting. I don't think it makes it better thats for sure. Maybe that is why they dont recommend rechargables due to less voltage sag under load. I'd throw in a 700 or 1050ma 7135 based driver in and use NiMh cells.
LED Driver - especially a one from DX - would add one more serious point of failure.
I'd add just a current limiting resistor there to secure from LED burn. I don't care about power effectiveness here - this is backup light. Backup light is not expected to be used during normal dive. And in case of **сkup it's expected to run for an hour or two without significant beam intensity decrease - enough to get out from there. I don't want the cheap LED driver to screw up the rescue plan.