Add sync bulk sandbox teardown for RLM env

[snimu]

Description

Use batched, synchronous Sandbox shutdown for vf.teardown.

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

---

Note

Implements synchronous, batched sandbox cleanup for RLM sandbox execution.

• Replace *_retained_sandboxes* with _active_sandboxes and track sandbox IDs on create/teardown paths
• On teardown, bulk delete _active_sandboxes via SandboxClient.bulk_delete in batches of 100; remove IDs from the set and log per-batch results
• Update cleanup flows to discard IDs after per-sandbox delete and to add session IDs to the active set for centralized teardown
• Add prime_sandboxes imports (SandboxClient, APIClient) to support sync bulk deletion

^{Written by Cursor Bugbot for commit 760e879. This will update automatically on new commits. Configure here.}