Implement Fuseki metadata store #106
This pull request implements a metadata cache implementation based on Apache Jena Fuseki to supplement the existing SleepyCat and SQLite implementations.
Fuseki can be run as a separate service to Gutenberg, e.g. via Docker, which makes setup of the library much easier: no more need to install bsddb3! This means that going forward we can move bsddb3 into an optional dependency. Additionally, Fuseki can be run on a separate machine from Gutenberg so it enables use-cases where multiple users may want to share a single metadata cache.