Robust list config parsing and improved SFT docs

[shralex]

• Added automatic string-to-list and tuple-to-list coercion in pyconfig.py for list[str] configuration fields (train_data_columns, eval_data_columns, trainable_parameters_mask, adamw_mask) to prevent unhelpful Pydantic validation errors.
• Added comprehensive unit tests in pyconfig_test.py covering 6 coercion scenarios and end-to-end integration.
• Added unit tests in instruction_data_processing_test.py validating that unsupported SFT columns throw helpful, descriptive assertion errors.
• Expanded docs/tutorials/posttraining/sft.md to document SFT schemas, custom chat templates, ShareGPT custom formatters, and linked a runnable example.

FIXES: b/520789228

Tests

Added unit tests to cover the new code.

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!