What it did is, in high mode it went off-on-off-on, timing was something like 1/2 second each, then maybe 5 seconds later it turned off. It never dropped down a mode, it stayed on high. Took 2 or 3 off-on clicks before it worked again.
I am measuring around 6.3A at the tail (though, that's possibly quite a bit lower than reality due to the meter), with either 3x 25Rs or MNKE 26650s. The cells were fresh off the charger and had run for I guess less than 3 minutes, cells were only down to 4.03-4.05v afterwards. It sounds sorta like the low voltage behavior, but neither mention anything about a total shutdown after the warning blinks.
The driver is dead; something I did to it while it was out of the light killed it. It was working, I separated the two boards to check something, and it never worked again.
I had a stock driver, with what I assume was the stock sense resistor setup of one 3.3ohm & one 0.5ohm. It did a miserable 2.3A at the tailcap. I added a single 0.24ohm with the other two still in place and tailcap current went to 7A (SEVEN!) with no signs of doing anything at all wrong (except making enough heat to burn holes in paper and blister your hand, etc.). The stock wimpy toroid is way small for this kind of current and it does get hot, I need to dig around and find something similar size/# of turns but with thicker or multiple strands.