Only really need to index the solr parts for relevancy purposes, not creating whole fedora objects or going through the rest of the ingest process.