What is your docs question about? Ask it
I assumed ShortestPathOneToMany used some clever algorithmic trick (like unidirectional Dijkstra is able to calculate all targets for one source in one go). Now, having read the algorithm source code, it seems to be almost the same as executing ShortestPathOneOne for all targets in sequence. The only difference I can see is that it is reusing the data structures across iterations. Is it only about reducing allocations or is there some deeper optimization that I cannot spot?

What do you suggest?
Add a section to README about how ShortestPathOneToMany works and why it is faster.

Additional context
None