code and tests to deal with mentions/credits

[ehenneken]

Code and tests (including data) for the implementation for mentions and credits.

Also: some documenting text on why the sorting of the data files has been changed in comparison to the sorting in the Classic index files (in the Classic Back Office).

[kelockhart]

@ehenneken is reference_count getting computed in here? citation_count is, and I've added reference_count to the protobuf, so just double check that reference_count is getting computed and sent along in the protobuf.

[ehenneken]

@kelockhart reference_count should get assigned in this block in process.py (which now also calculates mention_count and credit_count:

    def _add_citation_count_fields(self, return_value, original):
        author_count = len(original.get('author', ()))
        citation_count = len(return_value.get('citation', ()))
        return_value['citation_count'] = citation_count
        return_value['citation_count_norm'] = citation_count / float(max(author_count, 1))

I'd do that in a separate PR to keep things clean, but I can add it in this PR, if you prefer

[kelockhart]

@ehenneken either way! Just wanted to make sure it was on your to-do list for this pipeline.