Proposal: distributed workers and clusters

[abserari]

1. add a Registry, which could retrieve the bee info or code and dynamic use in beehive.
2. distribute, node as a worker that scheduled by beehive. only need connected to beehive master or admin node.
3. worker cluster, bee could have its own hive managed by themselves, when a request arrived, the beehive new bee with configuring or direct use one current bee to work.
4. well-defined for P2P network. everyone could run beehive subcommand to expose themselves service to beehive master that could be scheduler them. This design different from point 2.

[rocket357]

I could see this being particularly useful in a kubernetes environment where you could scale pods out based on workload.