Added caching functions for all query types except "fetch_all" if requests-cache is installed. Can turn request caching on with set_caching(), off with stop_caching(). The cache can be cleared with clear_cache().
Make "fetch_all" compatible with Elasticsearch v2.x scroll changes.
Added a "generator" parameter to getvariants (default False). When True, a generator to all results is returned rather than a list of all results. This requires much less memory for very large queries (eg vcf file annotation).