Refactor: Improve modularity and error handling in eval.py

[SyedaAnshrahGillani]

This pull request refactors the eval.py script to improve its structure, readability, and robustness.
The original script had a monolithic main function and lacked comprehensive error handling, making it
difficult to maintain and extend.

Key Improvements:

• Modularity: The main function has been broken down into two distinct functions:

  • load_config(config_file_path): Handles loading and validating the YAML configuration file.
  • evaluate_task(config): Contains the core logic for running the evaluation based on the loaded
    configuration.

• Error Handling: The error handling has been significantly improved:

  • The script now checks for and handles empty configuration files.
  • It gracefully handles FileNotFoundError if the eval_input_path does not exist.
  • It catches json.JSONDecodeError if the input file contains invalid JSON.
  • Error messages are now more specific and user-friendly.

• Readability and Maintainability: By separating concerns into smaller functions, the code is now
  cleaner, easier to understand, and more maintainable. This refactoring also establishes a clear
  structure for adding new evaluation tasks in the future.

These changes make the evaluation script more reliable and easier for developers to work with.