docs: playbook for unexpected service downtimes

[jku]

We should have a generic playbook for .e.g a longer GitHub Actions outage: probably no reason to go very detailed but would be good to have a general idea of what needs to happen to e.g. handle online signing with the KMS "manually"