docs: add AGENTS.md for AI coding agents

Add comprehensive documentation to help AI coding agents work effectively
with the envbox project. Covers:

- CRITICAL: Never push directly to main - always use feature branches
- Architecture overview (outer/inner container model)
- Critical security requirements (inner container must stay unprivileged)
- Development workflow and testing strategy
- Integration test emphasis and usage examples
- Common development tasks and patterns
- Package descriptions and responsibilities
- Coder integration details
- Common pitfalls and debugging tips

The guide prioritizes integration tests as the primary validation mechanism
and emphasizes maintaining the integrationtest package API.

Author: sreya