On the press the LA down with the bezel removed test, I think I forgot to mention that you need to turn on the tailcap. Either click it on if it's a clicky or twist it on if it is a twisty. Then when you press down on the LA to complete the circuit, the LA should light up if all components are working right. If the LA doesn't light, it isn't 100% certain that the problem isn't the bezel. It still could be the bezel as well as something else lower down. But to simplify things, I would assume that the bezel is ok, and there is just one problem, and it is lower down on the light (either the LA, batteries, or the tailcap).
On shorting across the battery and the exposed part of the SF battery tube with the tailcap removed test, you already kind of did this test when you tested for battery voltage. But for completeness, switch your meter to the high amps mode (typically 10A or 20A). Put the negative probe on the battery terminal. Put the other probe on the exposed end of the battery tube. If everything is working right, the LA should now light up. You should read about 1A of current draw if you have the P90 LA. If you have the P91, then you might get about 2.5A.
If the problem looks like the tailcap, first make sure all contacts are clean for both the tailcap and the exposed aluminum face on the battery tube. Put one probe from your meter on the tailcap spring. Put the other probe on the aluminum "saucer" part that you can see between the spring and the tailcap inner wall. That's the part that contacts the exposed aluminum of the battery tube when you press the tailcap button, completing the circuit. If the tailcap is a clicky, then click it on and off. When off, you should get infinite resistance. Clicked on, you should get continuity. For a twisty, ,wWhen you put your meter probes on the two areas described above, you should get continuity if the twisty tailcap is working right.
SF twisty tailcaps are very simple and use big chunks of metal, so it's hard to imagine that something has gone wrong with one. SF clickies are generally less robust (mainly under high amp loads), but still usually reliable.
From your test results, it does seem that your tailcap is the potential problem source, though. The one key test is to short across the battery and the battery tube to see if the LA lights up. If the LA lights up, then you know that your cells (and the LA, but you've already determined that) are good. Measuring resting voltage of 123A cells isn't 100% reliable in assessing battery condition (I assume by "good voltage", you measured around 9.0V-9.6V resting voltage).
Assuming that you aren't using an SF clicky like a Z49 or Z59, then I believe that the stock C3 tailcap is the same as the twisty, lockout tailcap on the SureFire 6P. CPF has posts on how to disassemble these tailcaps. Here is
one thread on the subject.
