FEAT: use pydantic response model for structured output

[gpernelle]

Directly Query for Pydantic Objects with GraphRAG

This PR introduces a powerful new feature that allows you to directly receive a Pydantic object as the result of a GraphRAG query. This significantly streamlines data handling and validation, enabling you to work with structured, type-hinted outputs immediately.

How to Use

By simply passing a Pydantic BaseModel as the response_model argument to the grag.query() method, GraphRAG will automatically parse and validate the query's output into an instance of your specified model.

Here's an example:

from typing import Annotated, List, Optional
from pydantic import BaseModel, Field
from fast_graphrag import GraphRAG

class GraphRAGTheme(BaseModel):
    """A single theme generated by GraphRAG."""
    theme_name: str = Field(description="Name of the theme")
    keywords: List[str] = Field(description="Keywords associated with the theme")

class GraphRAGThemeList(BaseModel):
    """List of themes generated by GraphRAG."""
    themes: List[GraphRAGTheme] = Field(description="List of generated themes")
    city_name: str = Field(description="Name of the city")
    total_themes: int = Field(description="Total number of themes generated")

query = 'your query....'
grag = GraphRAG(....)

# Directly get a Pydantic object from the query result
result: GraphRAGThemeList = grag.query(query, response_model=GraphRAGThemeList)

# You can now access your data with full type hinting and validation
print(result.city_name)
for theme in result.themes:
    print(theme.theme_name)