Every table in the Data Darbar warehouse, with its source, unit, caveats and columns — the reference for writing SQL in the console or reading the Parquet files from Python and R. Generated from the same catalog.json the console loads, so it cannot drift from the data.