While reading the through the event loop, it made more sense to me to inline the implementation - most of the actual apis are signature + docs, while the actual code is 2 lines each.

Co-authored-by: Mackenzie Zastrow <[email protected]>

* feat: add structured output support using Pydantic models

- Add  method to Agent class for handling structured outputs
- Create structured_output.py utility for converting Pydantic models to tool specs
- Improve error handling when extracting model_id from configuration
- Add integration tests to validate structured output functionality

* fix: import cleanups and unused vars

* feat: wip adding `structured_output` methods

* feat: wip added structured output to bedrock and anthropic

* feat: litellm structured output and some integ tests

* feat: all structured outputs working, tbd llama api

* feat: updated docstring

* fix: otel ci dep issue

* fix: remove unnecessary changes and comments

* feat: basic test WIP

* feat: better test coverage

* fix: remove unused fixture

* fix: resolve some comments

* fix: inline basemodel classes

* feat: update litellm, add checks

* fix: autoformatting issue

* feat: resolves comments

* fix: ollama skip tests, pyproject whitespace diffs

Related to #238

Co-authored-by: Mackenzie Zastrow <[email protected]>

Right now when you decorate a function you get back another function that mimics the behavior of a python module. Going forward, we're discussing adding helper/utility methods for AgentTool and in anticipatation of those changes return an AgentTool implementation that also acts like the original function.

Co-authored-by: Mackenzie Zastrow <[email protected]>