Implement ParquetFormatModel and update write_file to use the format API

[nssalian]

Continued work on #3100

PR Description

Follow-up to #3119. Implements ParquetFormatWriter and ParquetFormatModel, registers Parquet in the FileFormatFactory, and rewrites write_file to dispatch through the factory using the write.format.default table property. Future formats can be added in a similar way.

Rationale for this change

The write.format.default table property was never read - the write path was hardcoded to Parquet. This PR makes the property functional. Also threads file_format through _to_requested_schema / ArrowProjectionVisitor / _construct_field so field ID metadata keys are correct per format (PARQUET:field_id for Parquet, iceberg.id plus iceberg.required for ORC), preparing the write path for ORC support without changing default behavior.

Are these changes tested?

• tests/io/test_format_writers.py adds parametrized tests modeled after Java's BaseFormatModelTests covering round-trip, statistics, null handling, context manager caching, close idempotency, close-without-write, and ORC vs Parquet field ID dispatch.
• tests/io/test_pyarrow.py adds test_write_file_parquet_round_trip and test_write_file_dispatches_on_write_format_default exercising the full write_file path.

Are there any user-facing changes?

No. Default behavior is unchanged. Setting write.format.default to an unregistered format now raises a ValueError.

[nssalian]

@kevinjqliu @Fokko @geruh PTAL when you can

[Kurtiscwright]

Thank you for messaging me on slack about this PR. I only have one general question about the Python implementation.

Will the Python implementation support non-Arrow use cases?

For example Spark may want to use a custom columnar format, will that require translation from Spark to Arrow and then write to Parquet from Arrow?

The question the Rust File Format RFC is trying to answer:
Can the Spark format be extended to talk Iceberg specific semantics and then write to Parquet directly without ever needing to translate away from that Spark format?

[rambleraptor]

This looks awesome! I love how much it cleans this up. I've got a question, but otherwise I think this is good!

[rambleraptor]

Why not setup the writer here if it's set to None?