Add functionality to mark the tests as expected to fail

[martinobanana]

In order to backport tests to branches early and have them ready in correct time, as well as verify that the test is really failing on old rpms, we need a way of marking the tests as expected to fail.

When test is marked with something (tag?) as expected to fail, the test would run, fail but the overall result would be PASS, cleanly visible that something failed but was expected to fail.

There's a flag called xfail which was created for this purpose I believe but the behavior changed.