Automate error reporting for failed deployments

[shonfeder]

AFAIU, @mtelvers currently polls the build graph manually to monitor for failed image builds, and then triage the response.

It would save some cycles if we could automate this with A batch email or a slack message that tagged our team when builds failed, providing also a log excerpt giving us a sense for the problem.

I'm not sure if this is best done via graphana, notifications to #ocaml-org-deployer, or something else.