Do not ignore failure in Windows jobs due to timeout

What does this MR do?

This MR prevents Windows test job timeouts from being ignored. The reason we ignore job failures is because some tests are flaky. A timeout however should not be considered a success.

Why was this MR needed?

Assuming that a go test failure will cause go to exit with code 1, anything else should be considered a failure. I’m looking at https://gitlab.com/gitlab-org/security/gitlab-runner/-/pipelines/244643522 and if one notices the windows 2004 job, every single one of the parallel jobs timed out, but it is still listed as passed with warnings.

What's the best way to test this MR?

I've created a subsequent MR that uses this branch and shortens the Windows test jobs timeout to 9 minutes. The jobs that time out should be marked as failed in red, not in orange.

Example of job that fails on 9 minute timeout: https://gitlab.com/gitlab-org/gitlab-runner/-/jobs/1438651350 Similar job passing (notice the orange color of the failed badge): https://gitlab.com/gitlab-org/security/gitlab-runner/-/jobs/976240831

What are the relevant issue numbers?

Closes #27474 (closed)