Generalize notion of "backends" for specific kinds of deployments

Introduce a pluggable "backend" interface. Each instance of this interface contains logic specific for spinning up Cassandra instances in various environments. The existing logic would be moved to a "local" or "local install" backend. Additional backends might include a "docker" backend (for spinning up an instance using one of the defined Docker images) as well as vendor-maintained backends for the various hosted services (Astra, Amazon Keyspaces, etc.)

Main goal here is to make ccm a common interface for creating, managing and destroying instances across a wide range of environments.

No files, tests, or entry points are named. Start by locating the existing local cluster-management logic and map the operations that a pluggable backend must provide; done means that logic is represented by a local backend behind a common interface, with the scope for Docker and hosted-service backends clarified.

Written by the indexing model from the issue text.