Improve implementation of 'fail-on-impossible-test' test

In !20 (merged) a new test was added to the pipeline of this project. This test asserts the functionality of the failOnImpossibleTest configuration option.

The test asserts that a particular step fails. Its implementation is based on usage of allow_failure. This causes the test to issue a warning when it succeeds as intended.

We should get rid of this warning, as it suggests that something is wrong.

One approach is to add an additional parameter to our template component of something like "invertResult" that's only useful for testing & testability.

I am not sure if we should modify the artifact that we ship with so many changes only for testing purposes, even if it'd be just one parameter that only flips the result at the end. It'd be isolated, and you'd expect consumers to understand the need for testability, but it adds more complexity to something that's already growing to be quite complex. If we can keep the scripts of all runners consistent, that'd help for maintainability.