Prompted by migrating SonarQube to k8s mid-session, which briefly broke
Jenkins' connectivity to it (server URL config points at a fixed
IP:port that had to move). That's exactly the kind of outage this stage
shouldn't be able to turn into a failed deploy: analysis quality isn't a
deploy gate the way the build/Sonar-unrelated stages are.
Wrapped in catchError(buildResult: 'UNSTABLE', stageResult: 'FAILURE') —
if SonarQube is unreachable for any reason, the stage fails visibly (shows
red in the stage view) and the build is marked UNSTABLE (not hidden), but
the pipeline continues to Deploy instead of stopping there.