Because these tests restart nodes in standalone mode, it's possible for fast count, which doesn't discriminate between majority committed data and locally committed data and the true count, which only includes majority committed data on standalones to diverge. Therefore these tests should skip validating fast count.