Here are the photos as promised (and runtime graph), hope they help :]

I'm sure the GITD O-ring will reduce the output of my P3D XP-G R4 a little bit, but I can't measure by how much.
In the following beamshots, P3D R4 is always on the left and P3D R2 is on the right. The two lights were held by my hand ~1m from the wall.

There's no mistake in the image above, the P3D R2 actually has slightly brighter spill and also better throw with a tighter beam. Also, my P3D R4's beam isn't artifact-free. I measured the ceiling-bounce output again last night and here are some numbers:
-P3D R4: 8.92 lux (Throw = ~3000 lux @1m)
-P3D R2: 8.80 lux (Throw = ~4360 lux @1m)
-Eagletac P100C2 Q5:8.60 lux (Throw = ~8000 lux @1m measured at brightest spot)
-L-mini 2 R4 modded: 8.45 lux (Throw = ~3500 lux @1m)
However, all those lights have different beam profiles so the numbers didn't tell much.
***All of these data was obtained for personal use and I have no guarantee of their accuracy
For the runtime, I used a 120mm PC case fan to cool the P3D R4 so the light stayed cool throughout the test. The AW17670 battery was fully charged to 4.17V and rested ~3h.

Indeed, the P3D R4 stays regulated most of the time on 1x17670 battery. Its output increased slightly near the end of the test indicated the light was direct-driven, just for a couple of mins.