feat: improve infer_lang() robustness with shebang and shell pattern detection

[aditisingh02]

Why are these changes needed?

This PR addresses a TODO in autogen/code_utils.py to make the infer_lang() function more robust.

The previous implementation relied on a simple heuristic that only detected shell code if it started strictly with python , pip, or python3 . This missed many common shell scenarios and shebang-based scripts.

Improvements included in this PR:

• Shebang Detection: Added support for checking shebang lines (e.g., #!/bin/bash, #!/usr/bin/python) to accurately identify the language
• Expanded Shell Patterns: Significantly expanded the list of shell command patterns to include common commands like cd, ls, mkdir, git, npm, curl, wget, grep, etc.
• Robustness: Added handling for empty input and whitespace stripping before pattern matching
• Documentation: Added a proper docstring with Args and Returns

Related issue number

Note: This enhancement was originally proposed in the FLAML repository. As per @thinkall's guidance, raising it here since autogen has moved to this repository.

Checks

• I've included any doc changes needed for https://microsoft.github.io/autogen/. See https://github.com/microsoft/autogen/blob/main/CONTRIBUTING.md to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.