Is there a way to run a furthest neighbor query? Obviously the default query is doing nearest neighbor, and it would be poor performance to fetch all vectors and read from the end of the list (not to mention impossible until we get a larger max top_k)