Introduce caching for Document elaboration

[mwatts15]

A few of the external IDs (Pubmed, Wormbase, DOI) permit retrieval of external data. However, many of these services have limits (either declared or due to unstated resource limits on the server) on how frequently you can query them.

To reduce the #of times a given user will 'hit' a service, it's desirable to cache the data retrieved. The expiration date on such caches should typically be quite long by default (i.e., months), so cache size should be the dominant eviction criterion.