Allow a single central HNSW server

[bwhall61]

Right now it is possible for traversers distributed across many nodes of an HPC to talk to a single redis server that handles the traversal queue.

It would be good if we could do the same thing with the HNSW process. A single node which handles the querying of HNSW neighbors and just sends the appropriate data to the traversers to do the scoring. Right now each traverser needs to load a copy of the HNSW which can be memory intensive for larger graphs.