UV Refactor with Docker CPU

[alexnorell]

Key Changes

• Migrated from setup.py to pyproject.toml for modern Python package management
• Optimized Docker build process with multi-stage builds
• Consolidated and updated dependency management
• Removed redundant configuration files

Details

Package Management

• Replaced setup.py with pyproject.toml for package configuration and dependency management
• Introduced UV package manager for faster and more reliable dependency installation
• Removed .isort.cfg and pytest.ini in favor of consolidated config in pyproject.toml

Docker Optimization

• Optimized Dockerfile.onnx.cpu:
  • Switched to slim Python base image
  • Implemented multi-stage build to reduce final image size
  • Consolidated ENV variables
  • Updated to Python 3.13

Dependency Updates

• Updated and standardized dependency versions across requirements files
• Loosened version constraints on several packages while maintaining compatibility
• Dependencies are now properly categorized in pyproject.toml under optional features

Technical Notes

• Uses hatchling as the build backend
• Maintains all existing optional dependencies but organizes them better
• Dockerfile now uses UV for more efficient package installation

This change modernizes our Python package management approach and optimizes our container builds while maintaining all existing functionality.

---

Implementation Details

• Replaced multiple requirements.txt files with pyproject.toml optional dependencies
• Introduced multi-stage Docker build to optimize image size
• Consolidated isort and pytest configurations into pyproject.toml
• Updated dependency constraints to be more flexible while maintaining stability

[hansent]

@alexnorell @grzegorz-roboflow what it would take to get this across the line and switch to UV?