Adding a Return Format Checker Chain

[girishkc24]

Problem:
LLMs can be given a scenario and be asked to make a decision based on the provided facts. This is especially helpful in agent based simulations where LLMs can simulate individual agents. To be able to use LLMs in these use cases, it's critical that they provide response actions that are not only valid, but formatted correctly. Given the large context window of these scenarios, it's often the case that the LLM provides an invalid or incorrectly formatted decision.

Solution:
This chain helps format and validate actions that LLMs provide in response to situations. It does so by breaking up the prompting into multiple stages

1. Presenting the scenario and asking for the chosen action
2. Checking the validity of the action, and, if invalid, asking for a valid action
3. Formatting the action into the correct format
4. Checking the format, and, if invalid, asking for the properly formatted action

It's been pretty useful in making sure that the agents in these ABSs are producing good feedback without having to resort to writing more deterministic return format validators or try-except blocks.

[leo-gan]

@girishkc24 Hi , could you, please, resolve the merging issues and address the last comments (if needed)? After that ping me and I push this PR for the review. Thanks!

If this PR is not needed anymore, could you, please, let me know?

[GirishChandrasekar]

It's not needed anymore, thanks!

[leo-gan]

Closing.
@girishkc24 Thanks for your thoughts and effort! Really appreciated!