add pub fetcher as post-processing utility

[nlwashington]

in order to display the publications that any source contains with nice labels, it would be prudent to fetch the publication details from pubmed, if they are available. this could be done as a post-processing step like the following:

1. after the entire graph is built for a source, run a query to get all nodes that are publications
2. in batch, query using eutils for the basic publication information for any PMIDs. add the authors, title, short citation, year, and other publication metadata. consider adding the abstract, if we want to show it when hovering.
3. either insert the publication metadata back into the graph, or make a separate dump of the publication metadata into a new graph file.

also, this could be set to complete based on a commandline flag, because the time spent querying eutils might be quite high, depending on the source.

@cmungall do you like the idea of doing this at dipper time, or at golr time?

[cmungall]

It would be good to avoid re-querying each time. I think we'd want to maintain our own permanent cache. This could be SG, or another solution. If it's SG, would it be easy to merge the SG publication graph into each fresh monarch data graph?