I see the actual photons coming out of a light like this as one of the least important aspects for evaluation. One doesn't choose between a Sebenza and some Gerber folder based solely on specs. Construction and style matter a lot more.
I don't understand this stuff about crushing the cells, either. I'm all for fail-safe design, but I can't imagine what kind of user would crank down an Arc AAA enough to crush the cell. I do agree that Arc's choice of battery contact is unsatisfying, but the Fenix L1p's is not any better. I don't know what the E0 uses.
Finally, I also don't care much about flat regulation in this type of light. Go for maximum lumen-seconds integrated over the total runtime subject to some basic constraints on runtime; be able to operate at very low voltage with a long "tail"; and the heck with trying to keep the output perfectly flat. Arc has traditionally had a much more efficient circuit than Fenix, though maybe that's changed.
The Arc AAA and Fenix L1p are among my favorite lights though each has deficiencies that I think should be improved. If the E0 combines the best of both lights and adds its own improvements, that's great, but I'm sure I see that happening yet.