It's my understanding that the unregulated Battery Station (4.2 volts max) would run the LS in direct drive until the voltage drops to less than the Vf (forward voltage drop) of the particular LED. I heard it's generally not a good idea to run direct drive like that, as you might burn out the LED.
either burn out the LED or the circuit, whichever first. In any case, some folks have been successful in using R123 in their arc LS, modded or unmodded. It is generally not recommended, so do it at your own risk. though, YMMV
If you really want to explore all possibilities of making it work, you could put an ammeter in series with a fully charged R123 battery. #790 is a rev2 LSL if I recall and should be able to handle more heat than the rev1. If the ammeter is reading below say 600mA, stress to the LED will be not that bad. As a safety measure, let the light cool down if it gets hot in your hand during use. You could run it above 600mA, but the LED life will be shortened.
Also, the tempurature cutoff of the 1618 circuit used only has an effect in boost mode. Since the R123 is running the light in direct drive (passing straight through the inductor and diode, bypassing the effect of the 1618), the 1618's thermal cutoff does not help in this situation. That is why you need to be vigilent of the flashlights tempurature when using it. At higher currents, you will get less of a warning because of thermal delay.
The reason direct drive might work in some situations it that the Vf of the LED combined with the voltage drop of the schottky might be high enough to limit the current to safer levels.
Thanks shiftd and Peter. When I bought this light I had placed a .01" curved brass shim between the PCB and the housing to eliminate the 'flex-failure'. It's worked well every day since, but I might have reduced the rate of internal heat transfer to the housing, so... I think I'll continue to use it stock and keep my rechargeables for the TW4.