Features/feedback planner

[unsafecode]

Summary

This pull request includes various changes to improve the formatting, functionality, and structure of the codebase. The most important changes involve enhancing the vanilla_aiagents tests, updating the .vscode/settings.json file for better code formatting, and refining the PlannedTeam workflow to support reiterating a plan after feedback evaluation.

Enhancements to vanilla_aiagents tests:

• vanilla_aiagents/tests/test_llm.py: Added LLMConstraints and a new test for constraints to ensure the constraints are working correctly. Improved formatting and readability by adding line breaks and reformatting the code. [1] [2]
• vanilla_aiagents/tests/test_planned_team.py: Added imports for Conversation and improved formatting and readability by adding line breaks and reformatting the code. Introduced new tests for feedback and plan to enhance the testing coverage. [1] [2] [3] [4] [5]

Code formatting improvements:

• .vscode/settings.json: Configured the default formatter to ms-python.black-formatter and enabled formatOnSave for Python files. Added settings to organize and remove unused imports on save.

Refinements to PlannedTeam workflow:

• notebooks/planned_team.ipynb: Removed the stop_callback parameter from the PlannedTeam initialization to streamline the workflow setup.

Does this introduce a breaking change?

[X] Yes
[] No

PlannedTeam no longer accepts a stop_callback param - which was not very applicable. Instead a new repeat_until callback allows to reiterate and generate new plans enabling feedback-loop graphs.