Quite true ... and quite daunting from my end to go back and re-test all those older lights with 2600mAh cells. I have enough on my hands with all the new lights everyone keeps sending me.Selfbuilt, as noted above has only stayed with the 2200's to lend his review runtimes consistency so it's not fair to fault the factory for that.
Also, manufacturer's rarely use time to 50%, most use a relative "useable" light measure. If you pick the time my 18650s were almost dead, then by a simple calculation based on my results, a 2800mAh cell should give you a good 7 hours runtime on Med, and a 3000mAh cell would give you 7.5 hours (assuming everyone is calculating the mAh rating the same way - a big assumption). Taking into account emitter and battery variation, 8 hours is not such a stretch.
But I suggest it might also be possible that Olight has slightly changed the output from initial spec (i.e. 80 lumens seems a little low to me, I'm thinking it's more like 110-120 lumens, based on other lights I've tested). If the output of the final shipping version is indeed a bit higher than what they originally tested, than could also easily explain the rest of the variance.
Note that I'm not "defending" anyone's runtimes estimates - I actually believe manufacturer's estimates should be largely ignored, and the average of what individual users report be taken as the best indicator. In the real world, n=1 is not very reliable.