AnFI: The Blackhawk, for example, uses accelerometers to measure lateral acceleration. The signal gets fed into the AFCS, and account for control coupling, to include an input mixed with the stabilator when lateral acceleration is sensed (gust alleviation). (From old lecture notes, Fort Rucker origin, UH-60A)
(4) Provide sideslip to pitch coupling to reduce susceptibility to gusts. When the helicopter is out of trim in a slip or skid, pitch excursions are also induced as a result of the canted tail rotor and downwash on the stabilator. Lateral accelerometers sense this out of trim condition and signal the stabilator amplifiers to compensate for the pitch attitude change (called lateral to sideslip to pitch coupling). Nose left (right slip) results in the trailing edge programming down. Nose right produces the opposite stabilator reaction.
As to your speed estimate, OK, I see how you arrived at that. The last time I did a rotor over type maneuver was a few decades ago, so memory may not serve on how much speed is reduced going over the top and then accelerating ... maybe 90's closer than my guess.
Is your 90 kts the resultant vector in direction of travel, or is it already converted to X axis 90 kts and Y axis (some vertical speed) ? I ask because he picked up a bit of vertical velocity (acceleration even) after the turn over the top.