TestCase::setUpBeforeClass and TestCasetearDownAfterClass are not measured now and cannot be measured now, as currently the duration is measured per test.
What about measuring these too and reporting it as Cl::setUpBeforeClass/Cl::tearDownAfterClass items?
basic tests added in: #397
please note this must work with test process isolation correctly, where it might be counted into individual test duration
one example of (possibly) slow TestCase::setUpBeforeClass is https://github.com/atk4/ui/blob/5.0.0/tests/DemosHttpTest.php#L35, ie. things like starting (shared) webserver for a whole test class
related with #378, but not blocking