I have no idea what that means. But the knee point is just the point on the original PQ curve where the tone mapping curve begins. Values above the knee point are rolled-off or reduced in brightness. The shape of the tone curve determines how much values are rolled-off. It isn't really magic. Each luminance value is lower than the original by an increasing amount up to the display peak.
Dolby Vision would just provide formulas for different displays to reduce the overall dynamic range of the source for each scene based on a content flag indicating the peak brightness of the scene. It probably wouldn't be that different than what madVR is already doing. Tone mapping uses pretty standard formulas. It just needs to know how bright the input is in order to compress it.
This is what a 700 nits tone curve might look like:
https://i.postimg.cc/4NQWBHtC/PQ-Tone-Mapping-Curve.jpg
There is also an article here on tone curves:
https://www.insightmedia.info/should...-the-pq-curve/