Add a MapDB implementation of CitationCache

[sstults]

One of the limitations of our current in-memory cache is that the whole cache is stored in memory. This was an intentional design decision to prefer query speed over hardware cost. The one of the drawbacks of this approach is that each cache instance is only aware of the Lucene documents in the local index. In order to deploy Montysolr in a multi-shard collection we will need to implement a cache that uses the external document identifiers that are unique across the collection rather than the internal Lucene docids that are only unique to a shard.

This PR is a step in the direction of expanding the number of documents that can be indexed by reducing the memory requirements per-shard.