Add a more flexible grpc call wrapper

[philipjames44]

Per #22 (comment), we should add in a wrapper around gRPC connections/calls that:

1.) Adds in an ability to do optional backoff retries
2.) Keeps track of healthy gRPC endpoints on the fly and update cache

This PR should be done after #31. This is split into 2 issues to separate actual cache implementation and utilization bc the latter will require a bit of a refractor.

This PR should also extend the cache hashset into a hashmap with values indicating how many times in a row connection attempts have failed and remove endpoints that have failed more than some threshold value