Implement Structured Output with Pydantic Models

Add support for structured outputs using Pydantic models, similar to OpenAI's with_structured_output() method.

**Acceptance Criteria:**

- Add with_structured_output() method to ChatGradient
- Support Pydantic model validation
- Handle JSON parsing and validation errors gracefully
- Support both single and multiple structured outputs
- Add comprehensive tests for structured output functionality
- Update documentation with structured output examples

**Technical Requirements:**

- Use Pydantic for model validation
- Support both response_format parameter and with_structured_output() method
- Handle validation errors with clear error messages
- Maintain type safety

**Example Usage:**

from pydantic import BaseModel

class Person(BaseModel):
name: str
age: int
email: str

llm = ChatGradient(model="llama3.3-70b-instruct")
structured_llm = llm.with_structured_output(Person)
response = structured_llm.invoke("Create a person named John, age 30, email john@example.com")
# Returns: Person(name="John", age=30, email="john@example.com")

**References:**

[OpenAI Structured Output Documentation](https://python.langchain.com/docs/integrations/chat/openai/#structured-output)