ENH: automated issues on failures for bleeding-edge CI ? #3731
Description
A problem with our existing test setup, in particular regarding the bleeding-edge daily workflow, is that, as far as I understand, only the author of the workflow (me) gets notified when a job fails. Other maintainers may not be informed unless I report the issue manually, or they happen to notice that the badge in the README is red.
This has been the situation for a while now, but I just found out that there is a simple workaround this limitation in GitHub Actions.
Indeed it is possible to configure the workflow to open an issue in case of failure. See how xarray does it with their own "upstream dev CI" workflow (which is exactly the same idea as our "bleeding edge" job):
https://github.com/pydata/xarray/blob/main/.github/workflows/upstream-dev-ci.yaml
The one "cons" I can see is that adding this to our workflow increases ever so slightly our maintenance burden if it ever breaks, but this seems worth it. Any objection ?