[feat] Add Docker image publishing for easy adoption for everyone.

[marcelMaier]

Summary

• Add a CUDA-based Docker image for the SGLang batch inference path in infer.py
• Add a Docker Compose example for local NVIDIA GPU usage with mounted input, output, logs, and Hugging Face cache directories
• Add a GitHub Actions workflow that builds Docker images for PRs and publishes on main, version tags, or manual runs
• Document local Docker usage and the publishing secrets maintainers can configure

Publishing

The workflow publishes to GHCR with the built-in GITHUB_TOKEN. To also publish to Docker Hub, repository maintainers only need to add these repository secrets:

• DOCKERHUB_USERNAME
• DOCKERHUB_TOKEN

Validation

• docker compose config
• YAML parsed for .github/workflows/docker.yml and docker-compose.yml
• Verified Docker Hub contains the selected nvidia/cuda:12.9.1-cudnn-devel-ubuntu24.04 base image tag

[rajpratham1]

This is a comprehensive infrastructure improvement that adds first-class Docker support without changing the application's inference logic. The PR includes a Dockerfile, Docker Compose configuration, CI workflow for automatic image publishing, dependency management, ignore files, and documentation. Everything is scoped toward improving deployment and distribution rather than modifying runtime behavior.

[marcelMaier]

This is a comprehensive infrastructure improvement that adds first-class Docker support without changing the application's inference logic. The PR includes a Dockerfile, Docker Compose configuration, CI workflow for automatic image publishing, dependency management, ignore files, and documentation. Everything is scoped toward improving deployment and distribution rather than modifying runtime behavior.

Yes that was the goal to make it easy to adopt for everyone without much work.
I think this could give the project a nice boost.

[marcelMaier]

Superseded by #39.