Worker denial of service via recursive error handling

Uncontrolled recursion in Ash Framework AshOban 0.8.0-rc.1 through 0.8.13 allows local users to crash a worker via a failing on_error action. On a job's final attempt, the generated atomic handle_error/4 rescue calls handle_error/4 again with the unchanged job when the trigger's on_error action raises, re-entering the same clause without an exit. Exploitation requires an AshOban trigger with on_error and a repeatable action failure; each loop retains a formatted stack trace, consumes CPU and heap memory, and reissues the failing data-layer operation until the runtime kills the worker.