[Enhancement] Replace CSV-Based Benchmark Configuration with Pytest-Native Approach

[lcy-seso]

The project currently uses CSV files to define benchmark/test configurations. While functional, this approach introduces avoidable complexity and does not integrate well with modern Python testing practices.

This issue proposes migrating benchmark configuration to a pytest-native solution.

Problems with Current Approach

Using CSV for test or benchmark configuration has several drawbacks:

• CSV lacks type safety and structure
• Requires custom parsing and validation logic
• Poor integration with pytest (no native parametrization, markers, or filtering)
• Harder to review and document benchmark intent
• Limited support from tooling (linters, type checkers, IDEs)

Proposed Improvements

Preferred Option: Pytest Parametrization

• Define benchmark/test cases using pytest.mark.parametrize
• Keep configurations in Python for clarity, type safety, and tooling support

Optional Enhancement

• Evaluate pytest-benchmark for performance benchmarks to gain:
  • Statistical analysis
  • Regression detection
  • CI-friendly reporting

Acceptance Criteria

• Benchmark/test configurations no longer rely on CSV files
• Benchmarks are visible and runnable via pytest
• No custom CSV parsing logic is required for benchmarks
• Documentation explains how to add or modify benchmark cases

Scope Notes

• Initial migration can focus on correctness and clarity
• Performance tooling and stricter validation can be added incrementally