Summary

requestSessionEventRun currently logs and drops when triggerSessionEventRun throws. For transient failures (dispatch/model errors), queued exec system events may remain unprocessed until another unrelated inbound message/event triggers a run.

Why this matters

Exec completion wake is intended to produce an immediate follow-up turn. A transient failure should not silently lose that immediate-run attempt.

Scope

• Add retry/requeue semantics for failed session-event runs.
• Preserve serialization and avoid duplicate runs.
• Keep current hasSystemEvents guard behavior.

Suggested direction

• Requeue wake with bounded retry/backoff when dispatch fails.
• Include reason/source/session in logs and retry metadata.
• Add tests for transient failure then success, and exhausted retries.