Move skipping of transient failures to relate_failure_issue script and reenable reporting them in individual results

We had a lot of transient failure issues (namely related to Net::ReadTimeout) being created that weren't of much use. The solution we came up with was to not open new failure issues for these: https://gitlab.com/gitlab-org/quality/team-tasks/-/issues/1283.

!970 (merged) attempted to implement this change but I noticed on my pipeline triage rotation that there are still a fair number of failure issues being created with Net::ReadTimeout errors. It appears we stopped reporting the transient failures in the individual test cases and results issues instead of preventing new failure issues from being opened.

I propose moving the skipping of transient errors functionality from the report_results script to the relate_failure_issue script. This way we prevent new failure issues from being created for Net::ReadTimeout but still record these failures in the test cases and results issues.

/cc @mlapierre