fix: Fix type hint for output_schema parameter in ChatAgent class with Optional[Type[BaseModel]] and resolve the case of the complex Pydantic class

[Appointat]

Description

This Pull Request introduces enhanced type hints to the step and step_async method of our CamelAgent class, ensuring better type safety and clarity in its usage. The output_schema parameter previously lacked explicit typing, which led to potential misuse and confusion within our codebase and among module consumers. This update specifies that output_schema can optionally accept a type that is a subclass of BaseModel from Pydantic.

If the method is called with an instance of a BaseModel subclass (not the class itself), like this:

        from pydantic import BaseModel, Field

        class PersonaResponse(BaseModel):
            persona_name: str = Field(description="The name of the persona")
            persona_description: str = Field(
                description="The description of the persona"
            )

        class PersonaListResponse(BaseModel):
            personas: list[PersonaResponse] = Field(
                description="A list of related personas"
            )

MyPy will raise an error indicating that it expected a type (Type[BaseModel]), but got an instance instead. The error might look like:

error: Argument "output_schema" to "step" of "ChatAgent" has incompatible type "type[PersonaResponse]"; expected "BaseModel | None"  [arg-type]      
error: Argument "output_schema" to "step" of "ChatAgent" has incompatible type "type[PersonaListResponse]"; expected "BaseModel | None"  [arg-type]