[FEA] UMAP / TSNE should warn when knn_graph argument is invalid

[cjnolet]

While providing an intermediate workaround solution to a user in #1653, I made a mistake and wrote nearest_neighbors.kneighbors in a code example instead of nearest_neighbors.kneighbors_graph. However, this wasn't made immediately obvious to the user as they attempted to pass a 2-element tuple into UMAP using the knn_graph argument but it was ignored without any warning or failure.

There are a couple ways we could handle this. We should either directly support the 2-element distance/indices tuple as returned from kneighbors, we could fail, or we could warn that it's being ignored. I think there's benefits to either method, but I would be inclined to say perhaps we should consider trying to support the output as it is provided directly from kneighbors to make it easier for users and failing/warning+ignoring otherwise to make the behavior very clear.

[github-actions]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[github-actions]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.