Don't get me wrong, I think that AMC7135-based drivers are great for 1xLi-ion applications. I'm using them in several applications, including an LED turbo tower. If a 17mm diam driver is more convenient, and if price is important, then an AMC driver is definitely preferable to the GD1000. I presented the GD1000 to provide another choice which might be more suitable in size considering the desire to use a 14500 (AA-size) cell.
Regarding the XP-G vs MC-E, there are pros and cons to both.
The XP-G tint lottery IMO is iffy at best and I think that the chances of getting a non-greenish tint are low. Neutral and warm tints are still unavailable for the XP-G. On the other hand, the MC-E is more of a hassle to connect since you are probably looking at connecting its tiny terminals in 4P, if you work with a bare emitter. If you use an MC-E on a star, then you just buy the appropriately wired version.
An M-bin MC-E is rated at 430 lumens at 350mA per core (1.4A total). An XP-G R5 tested by jtr1962 put out about 440 lumens at 1.4A. Basically a wash. However, the XP-G is being overdriven at that drive current, while the MC-E is well within spec. Both have very low Vf, which is good.
I would expect that for the same reflector, the XP-G will have a brighter hot spot than the MC-E and thus better throw.
Underdriving an MC-E is a more expensive way to go. Even the DX/KD prices are in the ~$20 range, vs about $10 for an XP-G on an MCPCB.
But I wouldn't sell short an underdriven MC-E. My MC-E/GD1000 build runs very cool, with a very long run time. The LED draws only about 3W, yet delivers an estimated 300 lumens.