openclaw - 💡(How to fix) Fix Isolated cron task ledger rows can miss backing session keys [1 pull requests]

Problem

Isolated cron jobs can create detached task ledger records with no childSessionKey because task creation reads job.sessionKey. Isolated cron jobs normally derive their backing session from cron:<jobId> at execution time instead of storing job.sessionKey.

Impact

A successful isolated cron run can look detached from its backing cron session in task-status views or cleanup/reporting flows. Operators may see the task as missing its backing session even though the cron run completed successfully.

Expected

For isolated cron jobs without an explicit sessionKey, the task ledger should point at the stable backing cron session key: agent:<agentId>:cron:<jobId>. Per-run transcript sessions can still use the :run:<sessionId> suffix.

Repro shape

1. Configure an isolated cron job with sessionTarget: "isolated", payload.kind: "agentTurn", and no sessionKey.
2. Let the cron run successfully.
3. Inspect the detached task ledger row for cron:<jobId>:<startedAt>.

Expected: childSessionKey is the backing cron session key. Actual: childSessionKey can be missing.