Setup Ray Cluster for Contraction Hierarchy Runs

[wang-yinan]

Our example for Contraction Hierarchy has a Graph with 1189 nodes.

The dijkstra_with_contraction will need to be run around times just to form the list of all shortest paths.

• Setup Ray Cluster on AWS C5 instances (approximately 10 instances of C5.18xlarge).

Each instance provides 72 vCPUs, and we're able to break these CPUs down into half-CPUs using Ray.
This should allow us to setup a multiprocessing pool of around 1,440 concurrent workers, which should be able to process the entire dataset in around 17 minutes.

[wang-yinan]

The first todo will be to experiment with launching t2.micro instances using ray first. Should be done by the end of next week.