feat(example): add rl-training example

[hellomypastor]

Summary

• Added a new examples/rl-training example demonstrating RL training (CartPole + DQN) inside OpenSandbox, including dependency installation, training, checkpointing, and summary output. Also updated
  the examples index and aligned dependencies with the default opensandbox/code-interpreter:latest image. Resolves Add Reinforcement Learning (RL) Sandbox Example #29.

Testing

• Not run (example is environment-dependent; requires running OpenSandbox server and sandbox image)
• Unit tests
• Integration tests
• e2e / manual verification

Breaking Changes

• None
• Yes (describe impact and migration path)

Checklist

• Linked Issue or clearly described motivation
• Added/updated docs (if needed)
• Added/updated tests (if needed)
• Security impact considered
• Backward compatibility considered

[jwx0925]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Early returns leave sandbox running on failure paths

If pip bootstrap, dependency installation, or training fails, the function returns from inside the async with sandbox block before reaching the kill call at the end, while Sandbox.__aexit__ only closes client resources and does not terminate the remote instance. This leaves an orphaned sandbox running until an external timeout, consuming resources in failure scenarios; wrapping the block in a try/finally that always calls sandbox.kill() would avoid leaks.

Useful? React with 👍 / 👎.