Currently, we instantiate a new Parboiled parser for each native query execution. We should create the parser only once and have a stateless parser.