combine core and otherfeatures gene models

[ajo2995]

For maize we have the rejected set of gene models in the otherfeatures db and the filtered set in the core db. There is no overlap in the gene stable_ids in maize.
In triticum aestivum, for example, there are additional non-canonical transcripts stored in the otherfeatures database. These share the same gene stable ids as in the core db.

The gene dumping script should query both core and otherfeatures databases for a species so the set of models can be merged if necessary.

[ajo2995]

For now, we are going to explicitly ignore the triticum aestivum otherfeatures db. Previously, they were being ignored implicitly because of the way we were importing genes into mongodb (with mongoimport).

The otherfeatures models for maize and rice will be loaded (different stable_id's prevent conflicts). The db_type field in the solr docs can be used to distinguish the models from core and otherfeatures in the search result list or effect the order of the matching genes by boosting the score of core models over otherfeatures models. Also we can tweak the suggestions to prioritize core over otherfeatures