Currently, our full text search uses SQL's ilike which has to go through every entry. To be more efficient, we should make it use a full-text index.