You are going to have a driver in there somewhere aren't you? At 2S2P the LED will be facing 8 volts fresh off the charger, which is 30% over the 6 volt rating. At direct drive don't know what that would do, but it probably won't be nice.
Given a 700-800 mAh rating for the average 14500, at three amps you can expect sometime around a half hour of run time from this setup - 2S gives 6-8v and 2P gives around 1.5 Ah capacity. Also, three amps is about the limit for a protected cell, so you might want to look at some IMR/INR cells instead. If you already have some ICR 14500s you can try them, but don't be surprised if the PTC kicks in. I use Efest IMR 14500 cells (one at a time of course) in my EA11, which uses an XM-L2 LED, which is rated at a 3A max forward current.
As long as it is properly heatsinked it should handle a FET driver fine. Especially with 14500s which can't push a lot of current anyways.
Even most decent protected 14500's (Keeppower, AW, etc) can push 2A or more. With four of them, that's easily enough juice to provide the maximum rated power to an XHP50. He'll definitely want to ensure the MCPCB has a good thermal path to the body.
I seem to recall reading the TN4A battery holder is a tight fit with some batteries. Protected 14500's might not fit. Not a big deal, but that's something to look into before ordering batteries.
Fair point. I wasn't catching on that Thom might be planning to go beyond the official ratings. It will be interesting to see what the driver he found is able to push through the LED, especially if he uses some of the better 14500's.
Anyways, even the rated 18W is no trivial amount of heat get moved from the MCPCB to the body.