fix: detect the error twice

How the output is streamed changed a bit so we need to change the bats test

This will go back to zero if we ever re-implement how the container up starts for the docker-local scheduler.

tests/unit/ps-general-2.bats

@@ -111,7 +111,7 @@ teardown() {
   echo "output: $output"
   echo "status: $status"
   assert_failure
-  assert_output_contains "Could not start due to"
+  assert_output_contains "Could not start due to" 2
 
   run /bin/bash -c "dokku ps:set $TEST_APP procfile-path second.Procfile"
   echo "output: $output"