Beta: async streaming

[richardm-stripe]

Summary

Provides an implementation for request_stream_async for HTTPClientAsync, so e.g. await stripe.Quote.pdf_async(...) will work.

chunks = []
result = await stripe.Quote.pdf_async("qt_123")
async for chunk in result.stream:
    chunks.append(str(chunk, "utf-8"))

Details

• There's a new public type StripeStreamResponseAsync, which is the async analog of StripeStreamResponse. (This is the type that will be returned to the user when calling stripe.Quote.pdf_async). The difference from StripeStreamResponse is that the stream is represented by stream: AsyncIterable[bytes] instead of io: IOBase.
  • FYI: IOBase implements Iterable[bytes] -- so there's more symmetry between StripeStreamResponse and StripeResponse than appears at first glance.
• APIRequestor has been modified to use/produce it accordingly.
• The return type of HTTPClientAsync.request_stream_async is now Tuple[AsyncIterable[bytes], ...] and an implementation is provided for HTTPXHttpClient.

I was worried about a resource leak -- inside HTTPXHttpClient it calls response = httpx.ASyncClient().send(...) and then takes content = response.aiter_bytes() and only returns content to the caller. Does response get closed properly? I manually tested, and yes, after you do an async for through the bytes in content the response gets closed automatically.

[anniel-stripe]

LGTM! Left a few nits / questions but nothing blocking