Reddit user fnksb pointed out that the measured current, runtime, and expected battery capacity didn't quite add up for the 14500s. I tested both my Vapcell H10s on my analyzing charger and got 737 mAh and 797 mAh, which is considerably lower than the 924 mAh figure I had been using. I've recalculated efficiency numbers based on a battery capacity of 797 mAh and also reran the M1 mode runtime test.
I got a bit longer runtime this time around and of course a little higher efficiency in every mode. My overall analysis remains the same though: boost/linear is a little disappointing and I'd much rather see boost/buck or boost-only using a 6V LED like the Nichia B35AM. So far, I think only the Emisar D3AA takes the boost-only approach to the AA/14500 problem.