For what it's worth i believe the mechanism for the '92 anhedral tips is that tip vortex associated with producing lift is spead out (to make it weaker) and pushed away from the tip path. Helicopters produce lift by ingesting static air and accelerating it downwards through the rotor, so that downwash speed is twice speed through rotor. By pushing the tip vortex away the rotor effectively ingests air from 4% larger area than it otherwise would. So effectively disk loading goes down, or FM in the standard calc goes up.
I won't bore y'awl with the Biot-Savart law applied to vortex fields, and anyhow CFD has made this academic.
BERP I (or II?) was really designed for high speed, by having shaped tip profile stay within transonic Mach cone on the advancing side (sweepback), and incidentally pushing retreating stall up by letting the shed vortex developed during initial stall remain attached to profiled leading edge. Not sure it would make any difference in hover, but have seen no evidence to support or refute that claim.
I would be interested to see Nick's stall data, but my guess would be that once the already shed vortex is lost lift dropoff is severe.
BERP IV will likely be as much a step from BERP I as I was from 0.
Mart