HubLink POC — query Iceberg data from the browser

DuckDB-WASM reads the dataset's current Iceberg Parquet files, streamed same-origin from this app. union_by_name=true lets it read across a schema-evolution boundary (new column = null for older rows).