Update README.md

Update README.md

Point metadata config at parquet (CSV builder was being applied to alignment.parquet)

Add metadata.parquet so both viewer configs share the parquet builder

Update README.md

Fix alignment.parquet: performance_id as string, not Arrow dictionary (unblocks dataset viewer)

Add dataset card: schema, statistics, limitations, ethics, licence status

Upload 4 files

initial commit