I used to teach such a maneuver in the Cobra, it is doable and quite handy when in low level, high speed flight.
Mills is slightly off in two places:
1) where he says "I dumped collective... pulled the cyclic back.. and managed to cyclic climb to about 150 feet. With that extra altitude, I was able to enter autorotation..."
He was in autorotation when he started the cyclic flare, but with a combination of forward speed and rotor inertia, he was able to climb while in autorotation, and find a better place to touchdown.
2) where he estimates that he was at "about forty feet off the ground and doing maybe fifty knots".
The energy at 50 knots is paltry, and there is no kinetic energy to fuel such a climb. Were he at 130, the climb would be normal and nice, at 50, he wouldn't get more than a few feet up. The climb can be calculated easily, as the energy in his machine is 1/2MV squared (M is mass, V is speed), and the energy needed to climb is MGH (G is the gravitational constant, H is height). In a cyclic climb, the two must be equal (if the climb extracts all the energy, and you have zero knots at the top). For 130 knots, you get 750 feet higher, for 50 knots you get 110 feet higher. At zero, you can't autorotate, so you should calculate the climb with a residual speed of 40 knots for an OH-6.
At 130 knots, topping the climb at 60 knots allows you to climb almost 600 feet. At 50 knots, topping the climb at 40 knots allows you to climb about 40 feet.
Mills was wrong, or at least his memory was, but not by that much.
He was probably between 65 and 70 knots knots when he started the climb, and if he rounded out at 40 knots, he would have climbed about 150 feet.