fix: add AzureOpenAI class to guidance.models

[octo-patch]

Fixes #1205

Problem

guidance.models.AzureOpenAI does not exist as a public API, causing an AttributeError when users try to use it as documented in the AzureOpenAI example notebook. Users familiar with the guidance.models.OpenAI class pattern expect a similar AzureOpenAI class.

Solution

Add an AzureOpenAI class to guidance/models/_azureai.py that follows the same pattern as the existing OpenAI class:

• Takes model as the first positional argument (consistent with OpenAI)
• Requires azure_endpoint and azure_deployment arguments
• Supports api_version, api_key, azure_ad_token, azure_ad_token_provider, and other keyword arguments
• Automatically selects the appropriate interpreter based on model name (audio/image support detection mirrors create_azure_openai_model)
• Exports AzureOpenAI from guidance.models.__init__

The existing create_azure_openai_model factory function is preserved unchanged for backward compatibility.

Usage after this fix

from guidance import models

azureai_model = models.AzureOpenAI(
    model="gpt-4o-mini",
    azure_endpoint="https://my-resource.openai.azure.com/",
    azure_deployment="my-deployment",
    api_version="2024-02-15-preview",
    azure_ad_token_provider=token_provider,
)

Testing

Verified that:

• AzureOpenAI class is importable from guidance.models
• Syntax is correct (AST parse passes)
• Existing create_azure_openai_model tests remain unaffected
• The new class delegates to the same AzureOpenAIInterpreter classes as the factory function