slow1 tests have been failing recently with oom killer issues. Some of the slow1 tasks start up their own sharding test harnesses and replsettest harnesses. This puts excessive memory pressure on the test host. This is a temporary measure until we shuffle the tests around.