server_cluster_test.go

We can now run sharded queries on the server itself, simplifying the
client implementation. The server will forward the request to all
desired shards, it will store the data in a temporary SQL table, and
it will then run a *reduce* query on it to return the results to the
caller.