An existing reusable job must run as one stage within a larger workflow. Which task approach should be evaluated?

1. Copy only the job's display name into a log
2. Embed job output as a notebook Markdown cell
3. A nested job task
4. Replace every job with a new compute cluster