feat(python): expose FLATFILES on the PyO3 client

[userFRM]

Problem

sdks/python/src/lib.rs and historical_methods.rs have no FLATFILES bindings. Python users on the option-flatfile bundle have no way to pull a daily blob from the SDK.

Proposed solution

Add async methods on ThetaDataDx:

• await tdx.flatfile_request(sec_type, req_type, date, output_path, format)
• await tdx.flatfile_request_decoded(sec_type, req_type, date) — returns a list of typed row dicts. Add an Arrow path (to_arrow()) consistent with the existing tick story so Pandas/Polars work without round-tripping through CSV.
• await tdx.flatfile_request_raw(sec_type, req_type, date, output_path)
• The 9 convenience methods.

Expose SecType, ReqType, FlatFileFormat enums in enums_generated.rs (PyO3 enum). Type-stub them in the .pyi file.

Impact

Python is the largest installed base. This is the binding most users will reach for first.

[userFRM]

Closed by PR #514. Flatfiles SDK parity shipped via hand-written wrappers (commits 58deb80, abc265d, 269c16f, c81eea1):

• Python: tdx.flat_files.{quotes,trades,ohlc,...} returning FlatFileRowList with .to_arrow() / .to_polars() / .to_pandas() / .to_list() terminals
• TypeScript: tdx.flatFiles.* with .toArrowIpc() / .toJson() / .len()
• C ABI: tdx_flatfile_request_decoded, tdx_flatfile_rows_to_arrow_ipc, tdx_flatfile_request_to_path
• C++: tdx::FlatFiles namespace + tdx::FlatFileRowList RAII

Built on crates/thetadatadx/src/flatfiles/arrow.rs::rows_to_arrow (dynamic-schema Arrow) — hand-written rather than codegen because flatfiles columns are determined at runtime by (SecType, ReqType), which the static-schema codegen cannot represent. See #439 for that decision.