I believe the challenge that some vendors have in using NiMH batteries in AA style flash holders is the energy density that they pack.
I think it related to the current could get high enough, and the duty cycle high enough to heat the spring wire, though which the current flows such that it may loose some of it's spring characteristic.
In engineering terms, the metal would have yielded by the thermal exposure.
The other issue is the voltage output, as the flicker article discusses.
Since the original poster says that the thing works correctly sometimes, that would lead me to think that the problem is in the shut off circuitry, that feeds the power control tyristor. Thyristors like this don't usually go flaky, they just short closed.
The manual control, if it works, like for winder/motor drive modes, would show the thyristor is working. These modes switch in a fixed set of restistance to limit the shut off time to the thyristor.
The manual modes would also be a set of different ratio voltage divider pairs of resistors. I am basing this on a past opening of my Mecamat remote sensor to clean it after a mostly empty soft drink can dripped onto it. I will pack garbage separately next time!!!
I would not recommend taking it apart a Mecamat unless you have to. It is replete with a nest of resistors and contact pins and wiper arms that are a challenge to get safely back into their proper alignment to allow the case to close up.