Stabilize flaky timer and logging tests on slow database runners

The Oracle 23 CI matrix exposed five timer/async-job tests where a
fixed wall-clock budget left the AsyncExecutor too little time to
poll, lock and execute jobs after a mocked-clock advance. Bump the
budget to 10s where the existing assertion already gives a clear
failure diff, and switch to waitForJobExecutorOnCondition where the
helper would otherwise return successfully while the engine state is
still mid-flux (cyclic timer rescheduling, async log flushes after
the job is gone).