Add experiment tracking abstraction layer

[Sohailm25]

Introduces a unified tracking interface supporting multiple backends (Weights & Biases, MLFlow, TensorBoard, CSV) for experiment logging during training. Key features include:

• Abstract Tracker base class with standardized API for logging metrics, tables, and completions
• CompositeTracker for simultaneous logging to multiple backends
• NullTracker for disabling tracking without code changes
• Integration with GRPOTrainer via optional trackers parameter
• Comprehensive test coverage with 32 test cases
• Instance-level logging following codebase patterns

This provides researchers flexibility to switch between tracking backends and enables local development with CSV logging alongside production tracking with W&B or MLFlow.

Description

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Test improvement

Testing

• All existing tests pass when running uv run pytest locally.
• New tests have been added to cover the changes

Checklist

• My code follows the style guidelines of this project as outlined in AGENTS.md
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Additional Notes

[CLAassistant]

All committers have signed the CLA.