[Bug]: Healtcheck + completed workflows are broken in some edge case

[KevinMind]

What happened?

The workflow was run from the "schedule" event, but linked to a bunch of obscure Pull requests in different repositories https://api.github.com/repos/mozilla/addons-server/actions/runs/13907783652

• The healthcheck Dev job failed due to an un-handled network error requests.exceptions.ConnectionError which prevented the blocks json file from being created
• this lead the next step to fail before sending the slack message
• the completed job ran correctly, but it had malformed data so sent a weird looking slack notification

What did you expect to happen?

In this case we would expect the network error to be caught and handled, or that the completed job could run from a missing json file it seems to be sending incorrect data, maybe from running on a fork?

Is there an existing issue for this?

• I have searched the existing issues

┆Issue is synchronized with this Jira Task

[diox]

Need to investigate whether or not the network error was retried. We probably should have.