Use underlying SDKs to automatically fill in model options when creating drivers

[shhlife]

• [x] I have read and agree to the contributing guidelines.

Is your feature request related to a problem? Please describe. Customers can have a difficult time knowing what models to choose when creating drivers if they don't already know what models exist. For things like gpt-4o it's pretty easy to type.. but for options like sonnet unless you know the exact ids, it's a challenge.

It would be wonderful to be able to use the underlying model SDKs to grab the list of models available and dynamically fill them.

Describe the solution you'd like Here's a loom demonstrating a modification I did with Anthropic to pull from their list: https://www.loom.com/share/d7299383d6894101bc14db3fe8c05e43?sid=d4fedfee-ec64-45fa-9477-9a0a12dac9db

And to do the same with OpenAiChatPromptDriver I modified openai_chat_prompt_driver.py:

in the if TYPE_CHECKING: section:

if TYPE_CHECKING:
    from collections.abc import Iterator

    from openai.types.chat.chat_completion_chunk import ChoiceDelta
    from openai.types.chat.chat_completion_message import ChatCompletionMessage
    from openai.types.chat_model import ChatModel # Pull from OpenAi's types
    from griptape.tools import BaseTool

and then I switched the model param:

    # model: str = field(kw_only=True, metadata={"serializable": True})
    model: ChatModel = field(kw_only=True, metadata={"serializable": True}) # Use OpenAi's ChatModel

Describe alternatives you've considered We could manually create the list, but this seems way more efficient and will adjust automatically as the libraries change.

[shhlife]

I've been slowly implementing this in the Griptape ComfyUI nodes.. if it's helpful, here are the types for OpenAi:

from openai.types import AudioModel, ChatModel, EmbeddingModel, ImageModel
from openai.types.audio.speech_model import SpeechModel

[collindutter]

Blocked on #1587. Need to move imports out of TYPE_CHECKING to accomplish this.