Issues should be set to "needs work" when patches fail DrupalCI tests

Clarifying that both DrupalCI & QA.Drupal.org will set issues to "needs work". This makes the implementation more straightforward, and lets QA.Drupal.org catch any straggling DrupalCI false positives.

Normally, this might look like

1. Patch fails tests on DrupalCI, it is set to "needs work"
2. Patch fails tests on QA.Drupal.org, it is already at "needs work", nothing happens

The downside is that there is more chance for humans cross-posting with the robot, during the short window where both testing can update the issue status.

This will consider tests set to "Issues & on commit" for the branch as set in the project's "Automated testing" settings.

This is working well in my testing, and can be deployed later today.

This is now in production.