Answer what is actually in a bucket — list, preview and query Parquet/CSV/JSON on S3/R2/GCS with row counts and schema, without downloading anything
