Fix ruff-format failure in sandbox rollout observability changes

[Copilot]

The lint GitHub Actions job was failing on ruff-format in the sandbox rollout observability changes. This PR aligns the affected expression in sandbox.py with the repository’s enforced formatter output.

• Scope

  • Update the formatting of the sandbox rate-limit wait-time metadata assignment in xtuner/v1/rl/agent_loop/sandbox_agent_loop/sandbox.py
  • No behavioral or control-flow changes

• Change

  • Reflow the sandbox_acquire_rate_limit_wait_s accumulation expression to the single-line form required by Ruff

• Resulting code

record.metadata["sandbox_acquire_rate_limit_wait_s"] = (
    record.metadata.get("sandbox_acquire_rate_limit_wait_s", 0.0) + time.monotonic() - t_limit
)