Python: OpenAI Structured Output response_format ADR

[moonbox3]

Motivation and Context

This PR introduces the design for supporting OpenAI's Structured Outputs response_format within the Semantic Kernel Python SDK. While OpenAI’s previous JSON mode improved model reliability, it still lacked strict schema adherence, which led to inefficiencies and the need for custom workarounds.

By integrating Structured Outputs, this ADR outlines a design that ensures generated AI outputs will conform to predefined JSON schemas, improving consistency, reducing errors, and eliminating the need for retries or custom tools. This design will provide a more streamlined and reliable approach for developers building schema-compliant AI-driven applications.

In the future we will have a separate ADR around OpenAI's structured outputs for function calling. That is out of scope for this ADR.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the SK Contribution Guidelines and the pre-submission formatting script raises no violations
• All unit tests pass, and I have added new tests where possible
• I didn't break anyone 😄