[FEAT] Add support for Contextual Chunking from Anthropic

[bhavnicksm]

Hey there! 🦛

I want chonkie to support Contextual Retrieval that Anthropic talks about in their blog out of the box.

We can add the contextual retrieval from Anthropic via a ContextualRefinery class as a part of the Refinery module, such that given a particular generative module and original text, the ContextualRefinery can add appropriate context to the chunks, which can then be used for embeddings for vector search or in a bm25 index.

The API should look something like:

from chonkie import ContextualRefinery, TokenChunker, AnthropicGenie

genie = AnthropicGenie()
refinery = ContextualRefinery(genie)
chunker = TokenChunker()

text = ...

chunks = chunker(text) 
contextual_chunks = refinery(chunks) 

# use the chunks normally!

To get to this point, we need to complete the following:

• Add support for the Genie classes, specifically AnthropicGenie
• Plan for saving structured prompts for each genie inside the genie module
• Create the ContextualRefinery that can take in a genie module