Stricter handling of nf-test outputs

There are silent mishaps (i.e. non-failing test) in the nf-test output in nf-core/ampliseq that seem relevant for a pipeline to have fixed. But due to the test passing, those can be easily overlooked.

(1) in PR 1062, test 8/20 = '-profile test_multidb'

      > -[nf-core/ampliseq] Pipeline completed successfully-
      PASSED (785.039s)
    Snapshots:
      Obsolete snapshots can only be checked if all tests of a file are executed successful.

-> it seems to me that { assert !path("$outputDir/qiime2/barplot/level-8.csv").exists() }, refers to a not existing file leading to that message. But that seems important enough to me to report the test as failing.
Edit: As below might be a different problem (if the AI is correct).

(2) in e.g. (not the origin, because I didnt find it earlier) PR 1065, test 5/20 = '-profile test_glosed':

      > -[nf-core/ampliseq] Pipeline completed successfully-
  Error reading YAML file: /home/runner/_work/ampliseq/ampliseq/~/tests/a29655bff5567c226ec88c6ed9d3de74/output/pipeline_info/software_versions.yml (No such file or directory)
      PASSED (306.768s)

-> It appears that indeed that file isnt produced by the pipeline. I think that should result in a failing test.

I am suggesting to be stricter and make sure that those tests fail instead of pass. As a pipeline developer, I would like to catch theses issues without the need to dig into every log but instead trust the green check mark.

p.s.: Thanks for the useful tools!

Start by reproducing the referenced nf-test cases from PR 1062 and PR 1065 and inspect how their output is interpreted. Compare the missing snapshot and software_versions.yml messages with the reported green results; done means these missing or invalid outputs cause the relevant tests to fail visibly.