restart dead tunnels

[rasdani]

Description

monitors and restarts tunnels that have died in CliAgentEnv and RolloutGatewayEnv

depends on PrimeIntellect-ai/prime#403

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 tunnel lifecycle and rollout execution paths for CliAgentEnv/gateway mode; failures could surface as new TunnelErrors or change rollout timing/behavior if liveness detection is too aggressive.

Overview
Improves robustness of CliAgentEnv and RolloutGatewayMixin by detecting dead prime-tunnel processes, recreating them on demand, and surfacing mid-rollout tunnel failures as a new vf.TunnelError (including recent frpc output for debugging).

In gateway mode, adds a background tunnel health monitor that periodically restarts dead tunnels, tracks tunnel_id in state/logs, and ensures teardown_gateway() cancels the monitor before shutting down tunnels. Also bumps prime-tunnel/prime-sandboxes versions, increases gateway default timeout, and adds integration tests covering dead-tunnel recreation, health monitoring, teardown cancellation, and error raising.

^{Written by Cursor Bugbot for commit 3d9df75. 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.}