chore: reuse math rubric in hybrid math rubric

[mikasenghaas]

Description

HybridMathRubric now holds an internal MathRubric instance and delegates math_verify calls to it instead of running parse()/verify() itself. This reuses MathRubric's ThreadPoolExecutor-based, non-blocking verification with hard timeout and executor auto-scaling.

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

---

Note

Medium Risk
Changes core math-scoring behavior by routing HybridMathRubric through MathRubric’s executor-based verification and new timeout/worker settings, which may alter pass/fail outcomes and performance under load.

Overview
Hybrid math grading now reuses MathRubric for local rule-based verification instead of calling math_verify directly. HybridMathRubric constructs an internal MathRubric (with configurable timeout_seconds/max_workers), clears its auto-registered reward funcs, and delegates math_verify_score to MathRubric.correct_answer.

Minor cleanup includes removing direct math_verify imports/logging in HybridMathRubric and tightening docstrings/behavior around when the LLM judge fallback runs.

^{Written by Cursor Bugbot for commit a8e2704. This will update automatically on new commits. Configure here.}