Artemis: feat: Improve Token Estimator Documentation and Logic Clarity

[mike-turintech]

This pull request improves the TokenEstimator class by adding documentation and refactoring part of its logic.

Key changes:

• Added Javadoc comments to the public methods estimateTokenCount and estimateTokens, providing clear descriptions of their purpose, parameters, and return values.
• Refactored the estimateTokens method to calculate input, output, and total token counts into distinct local variables (inputTokens, outputTokens, totalTokens) before setting them on the QueryResult object. This makes the calculation steps more explicit and enhances readability.

The underlying token estimation logic remains the same.

Detailed Score Information

Score Details

This section contains detailed information about the performance scores for top 5 scored suggestions.

Top Performing Changes

1. src/main/java/com/llmproxy/service/llm/TokenEstimator.java:1-23 - Mean Improvement: 0.30, Mean Original Score: 4.51

• 🟢 Performance (Change: +0.07): The changes neither improve nor worsen runtime or memory usage. The only functional alteration is extracting repeated calls to estimateTokenCount for code clarity, but there is no noticeable impact on performance (as calling a lightweight getter/method twice for strings is trivial).

• 🟢 Quality (Score: 4.96; Change: +0.70): The code quality is significantly improved. JavaDoc comments have been added for both methods, clarifying their purpose, parameters, and return values. Additionally, redundant calls to estimateTokenCount are replaced by local variables, making the method easier to read and reducing the chance for subtle bugs or mismatches. This greatly increases code readability and maintainability.

• 🟢 Value (Change: +1.07): The changes add Javadoc comments to public methods, which enhances code documentation and understandability. Additionally, the introduction of local variables in the estimateTokens method slightly improves readability by making the calculation steps more explicit. While not fundamentally changing the functionality, these improvements contribute positively to the codebase's maintainability and clarity.