16 agentic walktrough with prompt optimization tutorial

[junkyul]

Added prompt optimization notebook in tutorial.
This case uses type manipulation and product/quotient while parallelizing the optimization loops.

[D3f0]

@junkyul do you think we can refactor to use Hugging Face's datasets to reduce the size of the PR

[junkyul]

@D3f0 I will check if the same data set is available in HF. Probably, there will be.

[D3f0]

@junkyul looks like the tain extended is this one: https://huggingface.co/datasets/openai/gsm8k
There are instructions, I wonder if we can implement from_dataset to pull data from huggingface.

[junkyul]

@D3f0
the dataset files are downloaded from HF. Those files are modified from the original dataset files so I created a public HF datasets junkyul/gsm8k.

[D3f0]

@junkyul should we try to consume your fork with datasets?