JSON_O1 Mode is outdated

[Pringled]

First of all, thanks for making this great library! While exploring the functionality, I encountered a bug.

What Model are you using?

• o1

Describe the bug
The JSON_O1 mode seems to be specifically designed as a workaround for the fact that o1 cannot handle system messages, streaming, and function calling. However, all of these features are now supported, which makes the JSON_O1 mode outdated and a bit confusing to use. See https://platform.openai.com/docs/models/o1 for an overview that states that o1 supports both streaming and function calling.

For system messages, the following example works:

from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
                    model="o1",
                    messages=[
                        {"role": "system", "content": "You are a helpful assistant who talks like a pirate."},
                        {"role": "user", "content": "What is the capital of France?"},
                    ],
                )

Which outputs "'Arrr, me heartie! The grand capital o’ France be Paris, the fair city o’ lights!'", clearly showing that the system message is used.

Proposal
Since deprecating the JSON_O1 mode might break functionality for users, I think simply calling handle_json_modes in future versions when JSON_O1 mode is used (and handle_json_o1 is called) would be the best solution, in combination with perhaps a warning message. If you agree, I'd be happy to submit a PR for this.