- Remove complex legacy launcher and replace with ultra-simple version
- Add --background flag for pipeline usage (starts vLLM, waits for ready, exits)
- Add health check support with configurable timeout
- Support CPU deployment with optimization parameters
- Clean project structure: single entry point, minimal dependencies
- Update package name to vllm-simple v1.0.0
- Remove setup.py, use pyproject.toml only
- Comprehensive README with examples and helper script documentation