feat: skills: support script execution

[kalenkevich]

Please ensure you have read the contribution guide before creating a pull request.

Link to Issue or Description of Change

2. Or, if no issue exists, describe the change:

Problem:
Skill Execution Limitations: Skills lacked the ability to execute scripts (JavaScript, Python, Shell) with input files and capture generated output files. There was also no mechanism to execute inline code generated directly by the LLM.

Solution:

1. Enhanced UnsafeLocalCodeExecutor:
   • Added support for writing input files to the temporary execution directory before script execution.
   • Added support for passing arguments to the executed scripts.
   • Implemented automatic detection and capture of output files created during execution, including mapping extensions to proper MIME types and encodings (e.g., JSON, CSV, images, PDFs).
2. New Skill Tools:
   • Added RunSkillScriptTool to execute scripts that are part of a skill package.
   • Added RunSkillInlineScriptTool to execute inline code provided by the model.
3. Comprehensive Testing:
   • Added unit tests for the new tools and the enhanced executor.
   • Added integration tests with mock responses to verify end-to-end execution for both JS and Shell scripts.

Testing Plan

Unit Tests:

• I have added or updated unit tests for my change.
• All unit tests pass locally.

Summary of passed results:

• Tests for UnsafeLocalCodeExecutor covering input/output handling and arguments.
• Tests for RunSkillScriptTool and RunSkillInlineScriptTool.

Manual End-to-End (E2E) Tests:

Integration tests serve as E2E verification for script execution:

• Verified JS script execution via integration tests in tests/integration/skills/script_js.
• Verified Shell script execution via integration tests in tests/integration/skills/script_sh.
• These tests verify that files generated by the scripts match expected outputs.

Checklist

• I have read the CONTRIBUTING.md document.
• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.
• I have added tests that prove my fix is effective or that my feature works.
• New and existing unit tests pass locally with my changes.
• I have manually tested my changes end-to-end.
• Any dependent changes have been merged and published in downstream modules. (N/A if self-contained)

Additional context

This PR significantly expands the capabilities of skills by allowing them to run complex scripts and produce artifacts.