Fix flaky test tests/token_billing/test_outbox.py::test_the_effect_runs_on_dbos_only_when_its_transaction_commits_and_a_failure_is_retried

The merge lane saw `tests/token_billing/test_outbox.py::test_the_effect_runs_on_dbos_only_when_its_transaction_commits_and_a_failure_is_retried` fail and then pass on a rerun of the same tree. - Test: `tests/token_billing/test_outbox.py::test_the_effect_runs_on_dbos_only_when_its_transaction_commits_and_a_failure_is_retried` - Tree: `3eb7b553cd0d19c9e8d7d803003c695b6ba1775e` - Lane task: 13264 - Flake record: 01a1a1d5-d1c9-7d2f-9f63-209721d7969c Make it deterministic; do not delete it unless it tests nothing.

Outbox live-engine test now waits for both DBOS workflows to finish before teardown, closing the race that crashed an xdist worker; terminal statuses single-homed. Quick gate and tests green, reviews passed; PR #7864 queued to the lane. Memory: none applied.