Only publish test results if triggering workflow was triggered by PR

[johannaengland]

Tested in my own private repo, both with a succeeding triggering workflow as well as a failing triggering workflow.

Reverts #2914.

Reference:
https://docs.github.com/en/webhooks/webhook-events-and-payloads#workflow_run (See: Properties of workflow_run)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 60.25%. Comparing base (1aec729) to head (e5d3f47).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #2922   +/-   ##
=======================================
  Coverage   60.25%   60.25%           
=======================================
  Files         602      602           
  Lines       44084    44084           
=======================================
  Hits        26565    26565           
  Misses      17519    17519           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lunkwill42]

Actually, I think we should maybe revert #2914 fully, and put #2862 on standby.

It seems that the test result publisher is indeed able to publish test results on the actions page when there is no PR associated with a commit. It's not that we want to avoid the test report, we want the reports - it's just that we thought there was nowhere to actually post it when there was no PR.

The confusing factor here is that the action fails, even if the results are published - apparently as it is trying to find a PR associated with the commit hash of the triggering commit.

Example in question: https://github.com/Uninett/nav/actions/runs/9063635622