[SPARK-55203][PYTHON] Support PathLike in readwriter paths

[201573]

What changes were proposed in this pull request?

This PR allows os.PathLike path objects, such as pathlib.Path, to be passed to PySpark readwriter path APIs.

The change normalizes path-like objects with os.fsdecode before sending paths to the JVM or Spark Connect plans.

Why are the changes needed?

Currently, several PySpark readwriter methods accept only str or list[str] paths. Python users commonly use pathlib.Path, and these objects should work for file-system backed data sources.

Closes #55203.

Does this PR introduce any user-facing change?

Yes. Users can pass pathlib.Path / os.PathLike objects to supported readwriter APIs.

How was this patch tested?

• ./dev/lint-python --compile
• git diff --check
• Added PySpark readwriter tests for pathlib.Path
• Added Spark Connect plan coverage for path-like path lists

Full PySpark runtime tests were not run locally because this machine does not have a Java Runtime installed.

Was this patch authored or co-authored using generative AI tooling?

Yes. I used OpenAI Codex to help implement and test this change. I have reviewed the changes and take responsibility for them. This contribution is my original work and I license the work to the project under the project's open source license.

[201573]

Additional local verification:\n\n- Built Spark with Hive profile: ./build/mvn -DskipTests -Phive package\n- Passed: pyspark.sql.tests.test_readwriter ReadwriterTests.test_pathlike_paths\n\nNote: I ran the PySpark test through a no-space local symlink for SPARK_HOME because my local checkout path contains spaces.