There have been new developments at the Map of Flashlights as to hotspot angles, including the near-final model description.
This is good news for the ray-tracing model.
For the special case of single smooth reflectors - if you know their specific h, D, and d values - a hotspot boundary derived from the ray-tracing model must coincide with some value of the max intensity fraction that I use to define where the hotspot ends, and thus potentially improve predictions for individual lights in this subset (as long as it complies with the etendue conservation).
A win-win situation!