Async-openai

- Update pricing to reflect OpenAI's new pricing model
- `gpt-3.5-turbo`
- `text-embedding-ada-002`
- Bugfix for consumption and usage validation in `chat` models
- Added support for `gpt-3.5-turbo-16k`
- Modified handling of `gpt-3.5-turbo`'s consumption pricing to reflect `prompt` and `completion` usage
- Modified default `Embedding` model to be `ada`

---

- Add better support for chatgpt models and `gpt-4`
- Better validation `max_tokens`

---

- Added support for GPT-3.5 Turbo through `async_openai.OpenAI.chat`
- Refactored `async_openai.OpenAI` to utilize a `metaclass` rather than initalizing directly

- Refactor `async_openai.types.options.OpenAIModel` to handle more robust parsing of model names.

- Fix proper charge for `babbage` and `ada` models.