test: feature_llmq_simplepose intermittent DKG/quorum timeouts under contention

[thepastaclaw]

Summary

feature_llmq_simplepose.py is intermittently timing out in DKG/quorum waits under contention. This showed up in dashpay/dash#7232, but it reproduces on plain develop, so it is not caused by that PR.

CI evidence

From dashpay/dash#7232 (ci: migrate primary CI jobs to ARM runners):

• failing check: linux64_tsan-test / Test source
• run: https://github.com/dashpay/dash/actions/runs/24068976186/job/70203705189

Failure in CI:

• test: feature_llmq_simplepose.py
• path: run_test() -> test_banning(self.force_old_mn_proto, 3)
• timeout: wait_for_quorum_list()
• symptom: after mining the final commitment, the expected quorum hash never appeared in quorum list within 60 seconds

Relevant excerpt:

AssertionError: Predicate ''''
        def wait_func():
            return quorum_hash in self.nodes[0].quorum('list')[llmq_type_name]
''' not true after 60.0 seconds

Local reproduction on develop

Checked out upstream/develop and ran:

python3 test/functional/feature_llmq_simplepose.py

This also failed locally on develop (outside the PR branch), with a different timeout in the same test:

• path: run_test() -> test_banning(self.close_mn_port) -> mine_quorum_less_checks()
• timeout: wait_for_quorum_commitment()
• symptom: minableCommitments never became visible on all nodes within 15 seconds

Relevant excerpt:

AssertionError: Predicate ''''
        def check_dkg_comitments():
            ...
''' not true after 15 seconds

Additional context

There is already related work in progress in PR #7254 (test: fix flaky PoSe ban assertion in feature_llmq_simplepose), but this current failure appears to be a broader timing-sensitive/simplepose flake: different wait points in the same test can fail depending on contention.

We also previously observed feature_llmq_simplepose failing under broad parallel contention while passing consistently when hammered in isolation, which points to resource-sensitive / inter-test timing behavior rather than a regression in the PR above.

Conclusion

This is a pre-existing flaky test on develop, not a regression introduced by dashpay/dash#7232.