Change API structure to account for a more general solution

[mpacer]

Just finished chatting with @willingc about the future of bookstore and how we would generalize the different pieces of functionality offered by bookstore to be less specifically tied to S3.

Carol's suggestion was to look into the papermill.iorw model and see how it could apply to bookstore.

There are some considerations that we need to think about that don't come up with papermill

1. As a server extension we need to think about how we want this functionality to be tied to the different routes
2. As a Jupyter server extension, we are using traitlets configuration systems to pass in S3 keys and our prefix conventions. We would need to provide some way to generalize that functionality as well.
   1. papermill does not address the configuration problem (as it is not a JupyterApp and does not integrate with the traitlets configuration system).

This relates to #110 indirectly.