Streaming response chunk size is incorrect about async-openai

[varon]

Thank you very much for this explanation and change.

[64bit]

Hi @varon

Thank you for detail explanation of the issue you were seeing,

async-openai emits the content as it is when available. The behavior you saw has to do with how example implements printing it to console. stdout is frequently line buffered by default, more about it here: https://doc.rust-lang.org/std/macro.print.html

So I have updated the chat-stream example #56 to make it visually similar to ChatGPT by flusing content to terminal when its available, here's the output https://raw.githubusercontent.com/64bit/async-openai/assets/chat-stream/output.svg

This issue will close automatically once I merge the PR #56, but please feel free to comment and/or open if you have any other issue.