Prepare release workflow, docs, and preprocessing utilities

[lmartinez2001]

Summary

This PR prepares the project for a cleaner release and onboarding flow.

Changes

• Add GitHub Actions workflow for CPU test checks
• Add GitHub Actions workflow for publishing a GHCR dev-base Docker image on version tags
• Expand README with devcontainer setup, data/ckpt mounts, checkpoint download, DONUT preprocessing, training, W&B logging, and release usage
• Add .dockerignore for cleaner Docker build contexts
• Update create_splits.py to optionally include tokens_path in generated manifests
• Update preprocessing smoke test for token-aware manifests
• Add/confirm dependencies needed for release utilities and logging

Testing

• Python source compilation passes
• GitHub Actions should run the full test suite on this PR

Notes

The Docker image is currently intended as a dev-base image, not a standalone runtime image.