Testing-system delegates test-running to test-runner. test-runner system can
be different from test environment. E.g. Ansible is run on the test-runner
and tests are executed on the managed host test . Usually for _test-runner is
used a stable version of OS.

The testing system SHOULD stage a clean test-runner for each set of tests it runs.

I would keep only standard-test-roles

. The testing system MUST stage the following packages on the test runner:

Thanks for the feedback, @astepano. I've updated the pull request.

@psss, thank you. Looks good.

Some tests, like the Cockpit CI tests, need to setup their own VMs for running tests. They launch qemu to do this. Is that compatible with these changes?

Yes, it should be compatible. Basically this is only clarification of the difference between test runner which runs tests and needs standard-test-roles to be installed and test environment where tests are executed. For Cockpit case I guess the VMs are test environment.

lgtm

LGTM also :clap:

Thanks for the feedback, merging.