Auto-update of versions: failures should be reported.

[chalin]

This issue offers some analysis of the failures that lead to a fix being proposed in the following PR:

• fix version update script #5758

After some investigation, here is what I've found:

(1) On Nov 13, 2024, 1:17 AM EST, the Auto-update versions #15240 job failed: https://github.com/open-telemetry/opentelemetry.io/actions/runs/11811741891/job/32905820377.

It failed, as @zeitlinger reports in #5758, because two runs of the subscript are attempting to create a PR for a single repo. The first subscript is successful, but the second fails because "a PR already exists" for the given repo version update.

I can't recall if maintainers get any notice about such failures? If not, maybe the script(s) should be altered so create an issue? Or we should have a process to monitor the GH actions and report any failures? Maybe @zeitlinger's proposal to have only one run per repo is the simplest solution.

(2) The next scheduled run at Nov 13, 2024, 2:16 AM EST of the script Auto-update versions #15241 passed because it saw that a PR had already been created and assumed that all was wel.

I think that we have an opportunity to improve things here. If there are changes pending, and a PR exists but has been merged, a new PR should be created. This would allow us to catch partial updates.

---

There is a separate problem that #5758 has surfaced, because it's build is currently failing, see https://github.com/open-telemetry/opentelemetry.io/actions/runs/12278556636/job/34260597540?pr=5758. For details, see:

• [CI/infra] Front-matter params: place them under the params key #5761