I have not seen it tested in a sphere yet but I have done ceiling bounce test with the RRT0 against several lights we know the true OTF lumens and my conclusions are this.
On a cr123 the RRT0 is about 180-190 OTF with a new battery, and it's not totally regulated on high meaning as the battery drains so does the lumen amount.
On 14500 it seems a bit brighter around the 200 lumen range and fully regulated until the battery is fully depleted.
Maybe mine is defective because I have done several test with the cr123 to just to check runtimes not thinking about output and the cr123 always dims when on max. On the medium mode I did one runtime and never noticed any dimming. I would love to see some runtime graphs and surprised there is none yet.
You take some of the best flashlight pics I have ever seen HKJ but I did not see a CR123 graph. Maybe I am reading it wrong but with the 16340 it looks like after the initial drop in the first minute it stays fairly well regulated until about the 31-32 minute.
I noticed you also show there the first generation RRT0 and I have the second, do you know if there was any changes in the driver in the newer one.