feat(data-types): Support PyArrow ExtensionTypes in pyarrow2athena

[richacode007-byte]

Feature or Bugfix

• Bugfix Support conversion from Pyarrow ExtensionTypes to Athena types #3346

Detail

• pyarrow2athena() in awswrangler/_data_types.py raised UnsupportedType: Unsupported Pyarrow type: extension<...> whenever it encountered any PyArrow ExtensionType. This made wr.s3.read_parquet_metadata() fail on parquet files containing canonical extension types such as pa.uuid() (and any user-defined extension type).
• Added a recursive branch in pyarrow2athena() that unwraps any PyArrow extension type to its underlying storage type, mirroring the existing pattern used for dictionary types. pa.uuid() now maps to binary (via its fixed_size_binary(16) storage), and user-defined extensions resolve to the Athena type of their storage.
• The check uses isinstance(dtype, getattr(pa, "BaseExtensionType", pa.ExtensionType)) to catch both canonical extension types (which subclass pa.BaseExtensionType in pyarrow 12+) and Python-defined extensions (subclasses of pa.ExtensionType), with a safe fallback on older pyarrow versions.
• Added three unit tests in tests/unit/test_s3_parquet.py covering: the UUID canonical extension, a custom user-defined extension, and athena_types_from_pyarrow_schema (the wrapper used by read_parquet_metadata).

Relates

• Closes Support conversion from Pyarrow ExtensionTypes to Athena types #3346

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[kukushking]

Thanks @richacode007-byte !

[linda-ting]

thank you! :~)

[richacode007-byte]

@kukushking @linda-ting Thanks for review and merging the code :)