KMedoids implementation

[adrinjalali]

scikit-learn/scikit-learn#11099 has a nice implementation of the algorithm, but it doesn't necessarily pass the inclusion criteria in sklearn, and would be a nice addition here.

@zdog234, would you be interested in porting it here?

---

From the original PR

Helping @Kornel to finish up/close #7694.
Reference Issues/PRs
What does this implement/fix? Explain your changes.

This is an attempt to finish the implementation of KMedoids clustering.
Any other comments?

Thanks to @Kornel and @terkkila for their work on this, and to everyone who's reviewed things so far. Also, thanks and apologies in advance to whoever reviews my code.

[znd4]

Just wanted to give an update. I'm finally getting around to working on this. @rth referenced a paper with an updated kmedoids algorithm. I need to read through it again, but it seems like it just uses a new heuristic for initial medoid selection, which should be pretty easy to implement.

[shalyng]

I just spent the last 10 minutes following the PR chain to here. Glad to see this continuing to get some love!

[rth]

Resolved in #12