Tonic support next steps

[gardnervickers]

#4 Brought in initial support for Tonic, but it is incomplete and there are several issues which makes it difficult to use.

1. @LucioFranco mentioned that we should provide our own transport for Tonic. I believe this entails providing an implementation of tower_service::Service which internally dispatches to the Environment handle for new connections.

2. To improve ergonomics a bit, it would be nice new Environment handles could be registered with a hostname, and the hostname be used to resolve network endpoints. I imaging something like:

   
   let server1 = runtime.handle("server1.cluster.local");
   let server2 = runtime.handle("server2.cluster.local");
   
   server1.spawn(start_server());
   
   server1.connect("server2.cluster.local").await;
   

3. The simulation-tonic bank.rs test currently uses timeouts to drop send/receive futures. This seems to result in occasional errors returned by Tonic.

   server error: grpc-status: Unknown, grpc-message: "http2 error: protocol error: unspecific protocol error detected
   

   It would be good to figure out how these can be better reported to users. Additionally, Tonic seems to support [automated reconnects] . (https://github.com/hyperium/tonic/blob/master/tonic/src/transport/service/reconnect.rs). Simulation users should be able to take advantage of this in their applications.