A Job with recovery.continuity: Restartable that reports progress with nodus.progress(completed, total) now
gets NODUS_CURSOR_COMPLETED and NODUS_CURSOR_TOTAL in its environment when it is recovered, so it can skip
the units it already finished.
NODUS_WORKLOAD_ID is set to the Job’s name, the same value as NODUS_JOB, so programs written for the
earlier Nodus keep working. It stays for one major version; use NODUS_JOB in new code.