[Feature] google collab A100 or base GPU exemple

[guidryheal-create]

Checklist

• This feature maintains backward compatibility with the current APIs in areal/api/
• If breaking changes are required, a refactor issue will be opened first

---

Background

In constrained environments (e.g. Colab, lightweight GPU instances, or restricted local setups), setting up AReaL can become difficult due to complex dependencies and system-level requirements.

While Docker is a common solution, it may introduce additional operational complexity in some cases (e.g. privileged containers, Docker socket passthrough, or nested container execution).

For this reason, a Google Colab-based execution/playground setup can provide a simpler and more accessible alternative for experimentation, learning, and quick iteration without requiring full system-level setup.

---

Proposed Solution

This PR introduces an example Colab-based workflow demonstrating how to run AReaL in a minimal, dependency-isolated environment.

Example notebook:
https://colab.research.google.com/drive/1Asub9UVkH_HtGOvVv1ZpuU5QKTbNVAHT?usp=sharing

Goals of this approach:

• Provide a zero-admin setup option for users
• Avoid system-level installation complexity
• Enable quick experimentation with AReaL pipelines
• Serve as a reproducible “playbook” for newcomers

---

Additional Notes

• A Docker-based setup would still be a valuable addition for production-grade or reproducible deployments.

• A future improvement could include:

  • docker-compose configuration for full stack orchestration
  • Prebuilt images for AReaL + dependencies
  • Optional Colab ↔ Docker parity for consistent environments

---

Summary

This PR focuses on improving accessibility by providing a lightweight Colab-based execution path, complementing (not replacing) existing Docker-based deployment strategies.