[Feature] Docker support

[brenocq]

Feature description
To simplify Atta setup across Windows, Ubuntu, and macOS - and to allow multiple versions to coexist - we should provide an official Docker image that’s published automatically on every new Git tag. Users can then spin up Atta via a docker-compose.yml (no local installation required), and have the Atta GUI launch with their project running. The image must support forwarding of graphics (Vulkan/OpenGL) and be extensible for GPU acceleration with CUDA in future releases.

Tasks

• Dockerfile & runtime setup

  • Create a multi-stage Dockerfile based on a lightweight Linux image, installing:
    • Atta runtime and dependencies
    • Vulkan/OpenGL loaders & utilities
    • A minimal X11/Wayland client environment for GUI forwarding
  • Configure environment variables and volume mounts to forward the host display (e.g. -e DISPLAY, -v /tmp/.X11-unix)
  • Add comments in the Dockerfile for future CUDA support (e.g. base on nvidia/cuda image when available)

• Example Docker Compose

  • Create docker-compose.yml example that:
    • Mounts the user’s project directory into the container
    • Sets up display forwarding for Linux hosts
    • Includes commented hints for Windows/macOS users (e.g. using Docker Desktop’s XQuartz/VcXsrv)
  • Add instructions in README.md on how to launch Atta via:

    docker-compose up --build

• GitHub Actions workflow

  • Add .github/workflows/docker-publish.yml triggered on push of any Git tag (refs/tags/*)
  • Use Docker Buildx to build multi-architecture images (linux/amd64, linux/arm64)
  • Authenticate to the chosen registry (GitHub Container Registry or Docker Hub) using GitHub Secrets
  • Build and push the image with the tag matching the Git tag:

    ghcr.io/your-org/atta:${{ github.ref_name }}

  • Add a job matrix for future CUDA-enabled variants (e.g. cuda vs. cpu)

• Documentation & testing

  • Document in CONTRIBUTING.md and README.md:
    • How to run Atta via Docker on each platform
    • How version tags map to container tags
    • How to extend for CUDA GPU acceleration
  • Test the workflow by creating a test tag and verifying:
    • Image builds correctly on GitHub Actions
    • GUI forwarding works on an Ubuntu host
    • The sample docker-compose.yml spins up the Atta window running a demo project