I doubt that 3xEneloops will be able to run in full regulation using that ShiningBeam 7135 driver unless your XM-L has exceptionally low Vf. At 2.8A drive, I'd expect Vf to be around 3.3V. If you look at the constant current discharge curves in the thread by Battery Guy "The AA NiMH Performance Test Thread", you'll see that Eneloops can delivery about 1.2V fairly steadily at 3A draw. So your Vbatt will be about 3.6V, against a Vf of 3.3V. But even if your Maglite has only 0.1 ohms parasitic resistance from the switch contacts, spring contacts, battery contacts, etc, you will lose ~0.3V and thus Vin is going to be at best equal to Vf, meaning that most likely you will either be in regulation only for a very brief amount of time, or not at all.
Also, you will be pulling about 10W from the 3xEneloops. If you look at Battery Guy's Ragone plot for Eneloops, they don't last very long at that constant power draw.
IMO, 3xEneloops is a poor choice.
For a low price build, I would go with those 2AA->D DX holders so that you can load your 3D Mag with 6xAA NiMH cells.
That will give you 7.2V nominal, which is plenty of extra voltage. Then get a cheap 2.8A or 3A buck driver and wire it up.
Edit: Actually, now that I look at those DX adapters again, it looks like they hold the 2xAAs in parallel? So you aren't going to get 7.2V for Vbatt. You are going to get 3.6V. It's hard to say if the 3S2P configuration will offset the probably extra parasitic resistance of those DX adapters. My guess is that it's still not that great of a choice for battery config. You really need more voltage for the 7135 driver.
If it were me, I'd find an FM or mdocod 9AA->3D holder on the Marketplace and use that with a low cost 3A buck driver. Otherwise, I'd probably take my chances with direct drive.
