^ following up my issue described above for those interested. I resolved the issue by using

result_storage="block_type/block_name"

(the storage block slug) instead of

result_storage=S3(...)

as described in the documentation.

@flow(task_runner=DaskTaskRunner, persist_result=True, 
      result_storage="block_type/block_name")

It now correctly persists results into S3. I wonder if this is an error in the documentation or I was incorrectly following/using the approach described in the docs.