There are four/five intermittently failing integration tests at the time of writing, tickets #15782, #15783, #15784 and #15785, which cause a lot of disruption to the stable branch tests (test_develop, plugin branches) and a little to pull requests.

I'd like to retry the failing integration tests when they fail to reduce the chance that the whole build will fail. This is preferable in my mind to skipping the test entirely, as it'll at least be run regularly still.