Expand Script Executor Python Module Support

# Bounty Points: 100
## PR Instructions
commits should be pushed to the *'script'* branch only

### Explanation of the Issue

The current Python script executor only supports the standard library, causing import errors when CoTAS generates code requiring popular data science libraries like matplotlib, seaborn, numpy, scipy, scikit-learn, statsmodels, and plotly. This restricts advanced analytics, visualizations, and machine learning capabilities.

### Maintainer's Notes

This enhancement is critical to unlock the full analytical power of the agent. Installing pinned versions of these libraries reduces security risks. Testing must confirm all packages import correctly and that generated analysis and visualizations run without errors. While containerization offers production-grade isolation, it adds complexity and potential compatibility issues.

### Resources or References

- Official documentation and user guides for matplotlib, seaborn, numpy, scipy, scikit-learn, plotly, and statsmodels
- Python virtual environments and best practices for `requirements.txt`
- Security considerations for running dynamic code with external dependencies

### Success Criteria

- Comprehensive `requirements.txt` including all key data science libraries
- Documentation updated with install instructions
- All packages install cleanly on supported Python versions
- Agent successfully executes scripts using these libraries and generates outputs like plots and models
- No regressions or breaking changes in existing functionality