Mitchaa wrote "The speed of rotation gains momentum which suggests (to me anyway) the yaw was fully extended with no way of it resetting/correcting. As it gained momentum it made the spin even more violent."
My speculation from experience of other designs; if the T/R control became disconnected, the blades would revert to a pre determined position by design. This position is a balance between aerodynamic and centrifugal turning moments which are usually designed to apply some positive pitch to the blades to allow a running landing. In a vertical climb OGE, this pitch setting would not be sufficient to prevent yaw developing (quite rapidly).
Now to address Mitchaa's observation of the rate of turn accelerating; as the yaw rate develops, the effective NR reduces, and more collective is required to maintain height, the torque reaction increases and yaw rate increases further, requiring increased collective pitch due to an even greater reduction in effective NR. It's a viscous circle. This would apply in the first few seconds before the descent was initiated. After descent is initiated, this theory falls down as the yaw rate would reduce noticeably and the video didn't really show this.
If the yaw servo did indeed travel to full deflection due to lack of feedback loop, then the rate of deflection, and consequent yaw, would be a function of how much the servo control valve was open at the time of failure. This could explain the relatively progressive increase in yaw rate initially.
If it was indeed a servo 'runaway', then deselection of the appropriate hydraulic channel should allow the blades to revert to the position mentioned above and permit some reduction in yaw rate and recovery of control.
I now refer you to a post I made
https://www.pprune.org/rotorheads/528810-category-takeoff-background.html#post10298782
There I tried to explain that tail rotors don't tend tofail 'quietly'. However, servo runaways are 'quiet' and should not be mis-diagnosed as a t/r drive failure, despite similar yaw rates.
JJ