Replace regex-based pytest parsing with a structured Python test runner

[NickK21]

Summary

Python evaluation currently parses pytest's human-readable terminal output with regex. This is brittle and makes error handling unreliable.

Current Problem

The Python path in codeval/executor.js uses parsePytestOutput() to scrape pytest terminal text.

This is fragile for cases like:

• assertion failures
• import errors
• syntax errors
• missing libraries
• no tests collected

Planned Changes

• Add a dedicated Python test runner that executes pytest and returns structured JSON.
• Update the Python execution path to consume that structured result instead of regex-parsing terminal output.
• Add regression coverage for Python evaluation edge cases.

Regression Cases

Add automated coverage for:

• normal pass
• assertion failure
• import error in program.py
• missing dependency
• syntax error
• no tests collected