CI: tests/test_closing.py::test_closing_higherfee address-in-use / timeout

[Lagrang3]

Flaky CI failure for tests/test_closing.py::test_closing_higherfee[False]

Observed in CI run: https://github.com/ElementsProject/lightning/actions/runs/30983919982/job/92236941520?pr=9380

Key log excerpts from the failing job:

• lightningd-2: BROKEN connectd: Failed to bind socket for 127.0.0.1:39061: Address already in use
• Time-out: can't find [re.compile('Server started with public key')] in logs (was previously present)
• Test: tests/test_closing.py::test_closing_higherfee[False] — TimeoutError / Node exited with return code 1
• Test summary: 1 failed, 85 passed, 9 skipped

Context and likely cause:

• The logs show an "address already in use" error for lightningd-2 and a subsequent timeout waiting for the node to announce its public key; this looks like an environment/port binding race or resource leak causing the node to fail to start reliably in CI.

Repro/CI metadata (from the job):

• repo: ElementsProject/lightning
• PR: [WIP] askrene-getroutes adaptive bounds #9380 (merge run)
• commit: 9ceb128
• run id: 343293042316947456

Suggested labels: flake, testing
Suggested priority: triage under CI flakiness; reproduce locally and/or on CI with increased logging and port isolation.

Action requested:

• Triage and confirm whether this is a recurring flaky test.
• If confirmed flaky, consider marking the test as flaky (pytest mark) or investigate port allocation to prevent address-in-use collisions in test environment.

Attached logs / evidence: link above (full job logs available on GitHub Actions).