Added datasets.CreateParquetReader

[janpfeifer]

This PR introduces datasets.CreateParquetReader to the datasets package.

Summary

• Added CreateParquetReader[T any](ds *Dataset, config, split string) (*parquet.GenericReader[T], error):
  • This function groups all Parquet files for a given config and split.
  • It downloads the files if necessary.
  • It creates a single reader for them by combining the row groups.
  • It also fixes the schema based on the generic type T.
• Updated ListFiles documentation to clarify the order of returned files.
• Updated docs/CHANGELOG.md.

This is useful when random access or more flexibility than sequential iteration is required.