Recent breakage with Windows shared runners?

My project flufl.lock is pretty stable so it doesn’t get many PRs these days. The last round of changes I made, added a Windows job which pulled all the active Python versions from NuGet, installed them and extended the path so that tox would work. Here is my previously working .gitlab-ci.yml file. However now I keep getting failures in the Preparing environment phase, and the link about shell profile loading gives me no help.

Here is a “debug” PR which shows that even with nothing changing in the .gitlab-ci.yml file, I’m seeing new failures. My project is using the shared Windows runners.

I’ve read over the GitLab CI pages, done lots of debugging PRs, search the forums and internet and cannot find anything related to these new failures. It seems to me that something changed in the Windows shared runner environment that caused Windows jobs to fail. I can’t find any more detailed logs to download to see exactly where things are failing. How can I further debug this problem? Or, is it a new infrastructure failure in the shared Windows runners?