feat(vector-search): enable Firestore vector search across sync and async models

[cto-new]

Summary: Introduce vector search capabilities to Firedantic, enabling nearest-neighbor queries against Firestore vector fields while preserving the library’s pydantic-centric design.

Details:

• Expose DistanceMeasure and Vector types in package exports for consumer model definitions.
• Extend VectorConfig and FiredanticVector to annotate model fields as vector fields.
• Implement vector_search on both AsyncBareModel and BareModel to return model instances by proximity.
• Discover vector fields automatically via _get_vector_fields and expose an Enum of field names (VectorFields).
• Support automatic resolution of the vector field when exactly one vector field exists; allow explicit field selection via a Union[str, Enum]; raise VectorFieldNotDefined or VectorFieldAmbiguous when appropriate.
• Use Firestore find_nearest with distance_measure and the vector field; support optional distance_result_field and transaction.
• Maintain backward-compatible API: new features are opt-in and do not modify existing methods.