Support all compression types for commonly used Parquet creation tools.

Polars enables compression types [including](https://docs.pola.rs/py-polars/html/reference/api/polars.DataFrame.write_parquet.html):
`{‘lz4’, ‘uncompressed’, ‘snappy’, ‘gzip’, ‘lzo’, ‘brotli’, ‘zstd’}`

Pandas enables compression types [including](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_parquet.html):
`None, ‘gzip’, ‘brotli’, ‘lz4’, ‘zstd’.`

Pyspark enables compression types [including](https://spark.apache.org/docs/3.1.3/api/python/reference/api/pyspark.sql.DataFrameWriter.parquet.html):
`(none, uncompressed, snappy, gzip, lzo, brotli, lz4, and zstd)`

Tested and working:
- uncompressed
- lz4
- snappy
- gzip
- zstd

Tested and not working:
- brotli

Untested:
- lzo

I believe this is determined by the libraries in the base image. Using Ubuntu vs Alpine solved some but not all.

Start by inspecting the container's base-image configuration and the compression behavior for the listed Parquet creation tools. Reproduce the reported working and failing cases, including brotli and lzo, and compare Ubuntu with Alpine. Done means all requested compression types work through the CLI.

Written by the indexing model from the issue text.