I appreciate your detailed analysis. My one thought is that I think your stats might be skewed by the fact that teams with the best PPs more frequently end the PP early (i.e. before the second unit has a chance to get out there). So those PPs will have a higher % of PP TOI given to their top forwards.
It’s interesting that even though Edmonton’s top forwards play the highest % of the PP time of any team, they actually have lower PP TOI per game than NJ’s top guns because they are ending PPs quickly.