DOC: Clarify behavior of sparse.csgraph.dijkstra for directed=False

[bdpedigo]

Issue with current documentation:

https://scipy.github.io/devdocs/reference/generated/scipy.sparse.csgraph.dijkstra.html#scipy.sparse.csgraph.dijkstra

I've used sparse.csgraph.dijkstra (and visited this page) many times; today was the first time I noticed the note which says:

As currently implemented, Dijkstra’s algorithm does not work for graphs with direction-dependent distances when directed == False. i.e., if csgraph[i,j] and csgraph[j,i] are not equal and both are nonzero, setting directed=False will not yield the correct result.

I find this behavior quite unexpected - I understand not wanting to do a check for symmetry and warn the user at every input to this function, but it still seems worth advertising this content more prominently.

if csgraph[i,j] and csgraph[j,i] are not equal and both are nonzero

Also, to clarify, does this imply that if csgraph[i,j] = x (x nonzero) and csgraph[j,i] = 0, then csgraph[j,i] = 0 is implicitly treated as having valuex if directed=False?

Idea or request for content:

I suggest putting some warning about this behavior under the directed kwarg description, even if it just points users to the Notes section for more information.

I would be happy to draft a PR for this if people support it.

Additional context (e.g. screenshots, GIFs)

No response