Add config option for fail-on-impossible-test

Test may be impossible to run, for example, because the server that's being tested does not support a particular feature.

It is not unreasonable for users to assume that, upon a successful test execution result, all tests have passed. In the case of 'impossible' tests, this isn't necessarily the case: some test might not have executed at all.

Especially in scenarios where users have configured the test run to enable a specific subset of the tests (through configuration such as enabledTests and enabledSpecifications) it may be desirable to hard fail when a test was impossible to execute. This enforces that a test run was strictly successful in all tests it was configured to execute.

fixes #5 (closed)

This also adds a test job for this new configuration option. The test job is is expected to fail. To prevent failure of the entire pipeline, allow_failure is set. This has the undesired side-effect of the pipeline to never go 'green' anymore. It, at best, can go 'orange': "success with warning"

An additional job is added that needs the prior job. It checks the status of that job (which is expected to be a failure). The status of that job is passed around in a dotenv variable.

This test approach is far from ideal (it particularly annoys me that we have to live with a 'warning'), but it's the best that I could do with my limited understanding of GitLab.