[hf] modify FQDNs for http servers
- due to a devops error (deleting some but not all images and volumes that the docker letsencrypt image depends on), we put letsencrypt in an inconsistent state, causing it to error on cert renewal attempts
- after enough failed attempts, it began to rate-limit, preventing us from being able to restore a consistent state (because requests from a well-formed docker configuration would still fail due to the rate limit)
- to work around this, we (temporarily) modify the FQDNs for
signalboost and signald http servers, the former of which is used
for
boostcli commands and scraping metrics, and the latter of which is used only for scraping metrics
as a side-effect, we add a tag to the provision_backup_src playbook
to make it easier to perform the rotation of deploy keys performed in !507 (merged)