Tests fail when they take longer
Some of the tests start to fail when they take too long or the test runner is starting with only a few or even just one worker(s). This seems to happen a lot with chromium, sometimes with firefox and never with webkit.