Handle endpoint connection errors

[rogeralsing]

Summary

• capture error returned from endpoint supervisor connection
• log failures and avoid dereferencing nil endpoint on connect

Testing

• make lint
• curl -s localhost:8500/v1/status/leader
• go test ./cluster/... -count=1
• go test ./scheduler -run TestNewTimerScheduler -v
• go test ./remote/... -count=1
• go test ./persistence/... -count=1

---

https://chatgpt.com/codex/tasks/task_e_68ac089d7bd483289bb79ac60d4537cd

[rogeralsing]

Thanks for tightening up the error path here. One thing I noticed: when returns an error, now returns without storing an endpoint. still does followed by , so we still panic—just a frame later. We either need to guard that load or store some sentinel before returning, otherwise the failure mode stays the same. Happy to re-test once that's adjusted.