Better support autopkgtests which fail due to increased testbed feature support

[Chris Halse Rogers]

There are a whole raft of SRUs that are stuck with autopkgtest failures on s390x. On inspection, these have all been cases where a test was previously skipped due to the test-bed not supporting machine-isolation that are now run because our s390x test-bed now supports VMs.

I've been marking such failures as “badtest foo/all/s390x”, but that seems like a hammer of unnecessary size.

*Ideally* brittney/the dep8 runner would notice “All the tests which used to pass still pass, but a test which used to be skipped is now run, but fails” and count it as a success.

Alternatively, being able to override part of the test-suite as badtest (badtest foo/machine-isolation/s390x or something).

Alternatively, being able to re-mark a test as always-failed.

[Chris Halse Rogers (raof)]

I don't know whether there's a LP project for brittney/our dep8 runner, so subscribing Colin and Steve, at least one of whom will *definitely* know!

[Steve Langasek (vorlon)]

This is a duplicate of LP: #1700668 or #1688516, depending on your view.