dev model agnostic structured output

[ohdearquant]

Why are these changes needed?

Related issue number

See #23 and please feel free to discuss the feature in there and the code development in here.

Checks

• I've included any doc changes needed for https://ag2ai.github.io/autogen/. See https://ag2ai.github.io/ag2/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[Hk669]

Hey @ohdearquant you can check on the new issue #23, feel free to propose new changes. Thanks

cc @marklysze

[qingyun-wu]

@ohdearquant let's work out a plan here. I would encourage to start from smaller PRs. Thank you!

[ohdearquant]

ok, @qingyun-wu I can make smaller PRs in future. Above requires a lot of architecture changes, maybe not the right move.

[ohdearquant]

@qingyun-wu @marklysze @davorrunje thoughts on making message into objects?

Why Object-Based Messages?

1. Enables more flexible operations on messages
2. Improves configuration and parameter management
3. Creates synergy with structured output patterns

Implementation Strategy

We could pursue a hybrid approach that balances structure and flexibility. For example, for structured output content formatting (per @marklysze's recent point), adding a custom format() to the Pydantic model is low overhead, @sternakt.

However, requiring specialized handling for every structured output can burden users. Not every output needs custom formatting, but conversation context benefits from flexible structure.

One potential solution: maintain the Pydantic structure while adding optional uniform formatting capabilities alongside the format() function.