Add SandboxMixin to CliAgentEnv

[rasdani]

Description

Extract sandbox lifecycle management into a reusable mixin for CliAgentEnv.
Can be used in SandboxEnv too.

Changes

• Add SandboxMixin class with ThreadedAsyncSandboxClient, retry logic, and bulk cleanup
• CliAgentEnv now inherits SandboxMixin alongside MultiTurnEnv
• Add active_sandboxes tracking to prevent orphaned sandboxes on Ctrl+C
• Add @vf.teardown hooks for bulk sandbox cleanup and client teardown
• Add retry with exponential backoff for sandbox API calls
• Use connection-pooled ThreadedAsyncSandboxClient instead of per-call AsyncSandboxClient

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Test improvement

Testing

• All existing tests pass when running uv run pytest locally.
• New tests have been added to cover the changes

Checklist

• My code follows the style guidelines of this project as outlined in AGENTS.md
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Additional Notes

---

Note

Medium Risk
Touches sandbox lifecycle and async concurrency/streaming paths (creation, teardown, retries, and cancellation), where subtle timing issues can cause leaks or hangs; changes are scoped to experimental envs but affect infrastructure behavior.

Overview
Refactors experimental sandboxed agent environments by introducing SandboxMixin (with a pooled ThreadedAsyncSandboxClient, retry/backoff, active sandbox tracking, and bulk teardown) and making CliAgentEnv inherit it; CliAgentEnv now centralizes sandbox creation/deletion, adds per-rollout timeout hooks, and adds extensive request/stream/LLM flow logging plus HTTP future timeouts to avoid hung intercept handlers.

Improves reliability around interrupts: MultiTurnEnv.rollout now explicitly runs cleanup on asyncio.CancelledError, and CliAgentEnv adds @vf.teardown hooks to bulk-delete leftover sandboxes and tear down the sandbox client threadpool.

Enhances HarborEnv with per-task timeout/resource overrides from task.toml, optional capture of per-episode artifacts from /logs, and more robust verifier execution (background-job based with tenacity retries) while skipping tests on prior infra errors.

Documentation updates clarify that @vf.cleanup methods should be idempotent and update CliAgentEnv parameter docs accordingly.

^{Written by Cursor Bugbot for commit 3e1ec44. This will update automatically on new commits. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}