Use proper wait in rndc test

Previously, the one-second sleep was unreliable, as it didn't properly
indicate that the rndc reconfig has been processed. The "test 'rndc
reconfig' with a broken config" check would sometimes fail under TSAN
in CI, because the previous rndc reconfig was still ongoing, and the
subsequent rndc reconfig was ignored.

bin/tests/system/enginepkcs11/tests_sh_enginepkcs11.py

@@ -51,6 +51,6 @@ pytestmark = pytest.mark.extra_artifacts(
 )
 
 
-@isctest.mark.flaky(max_runs=3)  # GL#4605
+@isctest.mark.flaky(max_runs=5)  # GL#4605
 def test_enginepkcs11(run_tests_sh):
     run_tests_sh()