If I am currently querying s3 parquet files with DuckDb, could I still do that with s3 tables, and what other advantages does this offer?