Fix networking issues within swarm

[ryardley]

Swarm is still not working correctly.

• Ensure kademlia is bootstrapping correctly
• Ensure nodes are sending messages to one another correctly
• Get nodes to accept multiaddrs that resolve their dns domains correctly (just about done)
• Enable nodes to not retry when dialing themselves
• Supply all nodes with multiaddrs of all other nodes
• Ensure we have a way to upgrade individual containers - Shutdown and update one service at a time
• Migrate all work on the example to enclave - possibly refactoring to an actor if easy
• Document everything

[ryardley]

Shifted work on this over to this repo which demonstrates exponential backoff within dialing to nodes: https://github.com/ryardley/libp2p-kad-gossipsub-quic-example/tree/main next step is to back port it to enclave

[ryardley]

TODO list for 19/12:

1. Finish getting nodes here to accept a multiaddr and swap out a resolved domain (close to done) https://github.com/ryardley/libp2p-kad-gossipsub-quic-example/blob/main/src/main.rs#L176

2. Handle what happens when a node is asked to dial itself cover the failure case there and don't retry it.

3. Supply bootstrap nodes with multiaddrs for all nodes in the cluster - this means that any node can be shut down and will automatically dial all the other nodes as it is restart - Q. do we need to persist kademlia routing table?

4. Fix deployment to shut down and update single nodes one at a time

   So far we have a script over here: https://github.com/ryardley/libp2p-kad-gossipsub-quic-example/blob/main/deploy.sh

   This shuts down the whole stack and restarts it again. This is so we can get the logs to appear to be from the newly deployed instances. For some reason I cannot update the instances and have the service logs respect it. Need a more detailed investigation of docker stack to work out how to manage this.

5. Migrate everything in the example repo to the enclave repo. This might involve refactoring the network peer to an actor