[Storage Backends] Implement storage backend APIs & persistent volumes

[JonathanMace]

This issue covers the broad concern of having services that we can migrate without losing their persistent storage. This mostly occurs in Kubernetes deployments, where we might want to migrate a service but keep its data.

This is not urgent, but it will be necessary eventually as some Blueprint use cases will clearly want persistent storage.

Implementing storage backends will include implementing a simple filesystem-like backend API, and backend options that include, e.g. HDFS/Cassandra/Zookeeper, as well as e.g. integrating with Kubernetes persistent storage.

The design of the above should be considered with care with respect to the representation and implementation of storage within Blueprint's IR, and the abstractions that get exposed to the workflow and wiring specs. We should discuss & design this before proceeding with any implementation.