Those commands are a (1) harder to grok and (2) do not actually use the memoized result (tmpfile.zst) to speed up a subsequent run. Can you give a more complet

Those commands are a (1) harder to grok and (2) do not actually use the memoized result (tmpfile.zst) to speed up a subsequent run. Can you give a more complete example of how you would use this to speed up developing a pipeline?