(Not a response, but related as an FYI: you can easily do sparse secondary b-tree indexes over secondary data using PostgreSQL using predicated functional indexes.)