I think Jason does a great job explaining some of the challenges in measuring the energy lost when the crankarm is deformed. He makes the very conservative assumption that all of the energy used to deform the crankarm is lost but allows that some of it may provide useful work later in the pedal stroke.
I think there is some confusion on when this “later in the pedal stroke” happens. Understandably some would think that this would happen on the upstroke and see little usefulness. The crank arm acts similarly to a very very firm springs and it springs back very quickly. Jason did not measure exactly how quickly and we can understand why, it happens continuously and very quickly as your leg reduces the force on the pedal during your downward stroke. Understand that the energy is not stored, held and then a tenth of a second later toward the bottom of the stroke (or worst on the upstroke) it is realized. As soon as the force on the pedal has reached a maximum, the crankarm is fully deflected. There is essentially no lag, and as soon as the force on the crankarm is reduced just a little the crankarm is now starting to release. Technically there is a lag, but my whole point is that it is so small that Jason cannot measure it. I suspect that it is something on the order of the speed of sound in aluminum or carbon fiber. Given the distance of deformation is less than a centimeter we can guess that the crankarms spring back quickly once the force is reduced from the maximum.
So while Jason has not measured the precise time it takes to release this energy, a good portion of that energy was stored very near the maximum pedal force and is will also be released very near the maximum pedal force. By the time the downward force on the pedal is significantly reduced, most of the energy that was stored in the pedal is no longer stored, it was either returned to the drive train or lost as heat. How much was lost as heat of course depends of the crankarm and while these things are difficult to measure precise we can have a good idea on “when” they happen.