README quickstart examples pass model/messages directly instead of the chatRequest wrapper

[slegarraga]

While reading through the README quickstart I noticed the chat.send() examples call send({ model, messages }), but the SDK's Chat.send() expects a single { chatRequest: { model, messages } } wrapper. Running the snippets as printed fails with a type error.

Suggested fix: wrap the payload in the README's non-streaming and streaming examples:

const completion = await client.chat.send({
  chatRequest: { model: "openai/gpt-5-mini", messages: [...] },
});

Happy to verify against the latest published version if useful.