Refactor IBenchmarkRegressionStrategy implementations to reduce code duplication through common variability analysis

[coderabbitai[bot]]

trafficstars

Problem

The qltysh static analysis tool has identified significant code duplication in the IBenchmarkRegressionStrategy implementations, with 88 lines of similar code across 2 locations (mass = 612).

Background

As part of PR #565, multiple regression detection strategies were introduced:

• P95RatioRegressionStrategy
• MeanPercentageRegressionStrategy
• MeanWallClockRegressionStrategy
• PercentageRegressionStrategy
• PercentileRegressionStrategy

These implementations share conceptual similarities in their approach to:

• Filtering regression results into 'better' and 'worse' categories
• Applying thresholds (both relative and absolute)
• Logging detailed performance information
• Computing geometric means
• Excluding infinite ratios

Proposed Solution

Apply common variability analysis to identify invariant concepts that can be elevated into a shared abstraction. This would:

1. Reduce code duplication
2. Improve maintainability
3. Ensure consistent behavior across strategies
4. Make it easier to add new regression detection strategies

Acceptance Criteria

• [ ] Analyze commonalities across all IBenchmarkRegressionStrategy implementations
• [ ] Design and implement a base class or helper that captures invariant behavior
• [ ] Refactor existing strategies to use the common abstraction
• [ ] Ensure no regression in functionality
• [ ] Update tests to cover the refactored implementation

References

• Original PR: https://github.com/rjmurillo/moq.analyzers/pull/565
• Discussion: https://github.com/rjmurillo/moq.analyzers/pull/565#discussion_r2217992922
• Static analysis finding: https://qlty.sh/gh/rjmurillo/projects/moq.analyzers/issues/qlty/similar-code/issues/850d82b419fc55e5568f994a2ca69514