Add OpenAI processing example

[tillrohrmann]

Scenario: pipe a larger CSV file through OpenAI API to create embeddings to store somewhere. Basically a mini batch-workflow across external systems. This is common for other engines (airflow, etc.) but would be great to show how simple that also is in Restate.

• Request -> pull (stream in) CSV file (maybe from S3, whereever) in a regular handler
• file chunked (chunks of 100 rows) and sent to a worker handler, but in parallel (rpc, but only collect promise, don't await)
• worker handler loops over rows and makes side-effect call to API, then store in DB.