I don't know that much about Luxeons (although I absolutely love my FF w/ MM+ and TY0J lux III) but a few theoretical considerations come to mind. The internal resistance of a 123 is something like 300 mOhms which really limits the theoretical power able to be delivered to a load. The voltage across the terminals would be 3 volts - .3 * i. Thus the power delivered to the load would be 3*i - .3*i^2. To find the maximum take the first derivative and set it to zero: 0 = 3 - .6* i or i = 5 amps. At 5 amps the voltage would be 1.5 volts, and thus the power would be 1.5 volts * 5 amps = 7.5 watts.
OK. So 7.5 watts. Now, so far, the maximum efficiency I have seen is around 30 lumens per watt of bulb lumens. Remeber that what finally comes out the front end will always be less than this. 75 percent is a good round estimate of what makes it out the front. Anyway, 30 * 7.5 *.75= 169 lumens. But not for very long, for sure, if it could even really support a 5 amp draw. SureFire applications seem to indicate that 5 watts per 123 is more or less the most that can be expected, so that would translate into 112 lumens out the front of the torch.
123 cells have built in thermal protection, however, and would probably shut down in a matter of minutes or even seconds when asked to provide this kind of power.
Still, there you have the theoretical maximum, assuming that the true voltage source (=o.c. voltage) in series with a resistor (=internal resistance) is a good model for a battery over all current values. This is not the case, however, as you have to consider electrolyte depletion and other groovy stuff like that which is really beyond my knowledge at the moment.
Eh, take it for what it's worth. Hope this helped.