Capture runner death logs before pods are reaped

The runners controller now captures the runner container’s log tail when a runner pod ends abnormally mid-job, and re-emits it to the controller’s own stdout before the pod is reaped. This preserves the diagnostic trail that was previously lost when the pod was deleted, so operators can investigate ‘self-hosted runner lost communication’ failures. The Helm chart grants the required pods/log permission on the controller Role. After deploying, search Loki for the message ‘runner death log captured’.