Preserve column order in generated orm.yaml

[yhong123]

Summary

• yaml.dump in make_tables_file defaulted to sort_keys=True, alphabetising all dict keys in the generated orm.yaml — including each table's columns.
• This ordering propagated downstream through dict_to_metadata/dict_to_table, causing exported CSV/parquet column order to not match the source parquet/database schema.
• Pass sort_keys=False to yaml.dump so the original column order (already correctly preserved by parquet2orm.py) is retained through to the exported files.

Fixes #135

Test plan

• Generate synthetic data from a parquet file with a known non-alphabetical column order and confirm the exported CSV/parquet columns match the source order.

[tim-band]

Great