[FLINK-40418][python] Support attribute-based column access on DataFrame

[beetle0915]

What is the purpose of the change

Implement FLINK-40418, the attribute-based column access described in FLIP-591. This allows df.name to return a column expression when the name is a valid Python identifier, is not a keyword, and does not conflict with an existing DataFrame attribute.

The implementation follows the approach discussed and accepted on the Jira issue.

Brief change log

• Add DataFrame.__getattr__, preserving normal Python attribute lookup and reusing the existing Table column expression API.
• Raise AttributeError for invalid or missing column names; guard against recursive lookup on uninitialized instances. Column lookup resolves the current schema without executing a Flink job.
• Add unit and execution-level tests, and document attribute access and bracket-access fallbacks.

Verifying this change

This change adds 13 tests covering valid and missing columns, identifier rules, attribute conflicts, transformed schemas, uninitialized instances, schema errors, and execution of a filter/projection using attribute references.

Local validation on Python 3.12 and Java 17:

• python -m pytest flink-python/pyflink/dataframe/tests -q: 352 passed (10 deprecation warnings).
• From flink-python: python -m flake8 --config=tox.ini pyflink/dataframe: passed.
• From flink-python: python -m mypy --config-file tox.ini: passed (83 source files).
• Full Python HTML documentation build with the repository-pinned Sphinx dependencies, -a -W --keep-going: passed with no warnings.
• git diff --check: passed.

The local validation does not include the full repository mvn clean verify, the full end-to-end suite, or the other supported Python versions; these checks remain to be covered by CI/review.

Does this pull request potentially affect one of the following parts:

• Dependencies (does it add or upgrade a dependency): no
• The public API, i.e., is any changed class annotated with @Public(Evolving): yes
• The serializers: no
• The runtime per-record code paths (performance sensitive): no
• Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: no
• The S3 file system connector: no

Documentation

• Does this pull request introduce a new feature? yes
• If yes, how is the feature documented? Python API reference and method docstring, including examples and fallback syntax for conflicting or invalid names.

---

Was generative AI tooling used to co-author this PR?

• Yes (please specify the tool below)

Generated-by: OpenAI Codex (codex-cli 0.154.0-alpha.6.2)

[flinkbot]

CI report:

• 055a478 Azure: SUCCESS

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build