Add LLMLingua Prompt Compressor in DSPy

[mango-panther]

Related to issue #13
This PR adds Prompt Compressor using the LLMLingua library in DSPy
It introduces new PromptCompressor class which uses the functionality from LLMLingua

Below is the code snipper that I have used to check whether working of code:

import dspy

class CheckCitationFaithfulness(dspy.Signature):
    """Verify that the text is based on the provided context."""

    context: str = dspy.InputField(desc="facts here are assumed to be true", compression=True)
    text: str = dspy.InputField(compression=False)
    faithfulness: bool = dspy.OutputField()
    evidence: list[str] = dspy.OutputField(desc="List of exact quotes from the text supporting evidence for claims")



def main():

    # create compressor 
    compressor = dspy.PromptCompressor(
        model_name="microsoft/llmlingua-2-bert-base-multilingual-cased-meetingbank",
        dynamic_context_compression_ratio=0.55,           # desired compressed size as fraction of original tokens
        preserve_fields=["username, history"],        # list of regex/keys to preserve fully (e.g., user names)
        target_token=None,      # hard limit to compressed result if desired
        device="cpu",                # or "cuda"
        cache=True                   # cache results for identical inputs
    )
    lm = dspy.LM("ollama_chat/llama3.2:1b", api_base="http://localhost:11434", api_key=api_key)
    dspy.settings.configure(lm=lm, compressor=compressor)
    evaluator = dspy.Predict(CheckCitationFaithfulness)
    long_document_context = (
        "Belgium, officially the Kingdom of Belgium, is a country in Northwestern Europe. "
        "The country is bordered by the Netherlands to the north, Germany to the east, "
        "Luxembourg to the southeast, France to the southwest, and the North Sea to the "
        "northwest. It covers an area of 30,689 square kilometers and has a population "
        "of more than 11.5 million. The capital and largest city is Brussels."
    )
    
    # 5. Execute the prompt!
    result = evaluator(
        context=long_document_context,
        text="Belgium's capital is Brussels and it neighbors the Netherlands."
    )

    print(result.faithfulness)
    print(result.evidence)
    print(result.compression_metadata)

if __name__ == "__main__":
    main()

Its output is:

True
["Belgium's capital is Brussels and it neighbors the Netherlands.", 'The Kingdom of Belgium has its capital in Brussels.']
{'total_original_tokens': 82, 'total_compressed_tokens': 37, 'ratio': 0.45121951219512196, 'model': 'microsoft/llmlingua-2-bert-base-multilingual-cased-meetingbank'}

I have also updated docs, and checked them

[mango-panther]

I dont know what to do with cache=True as LLMLingua doesn't have that particular parameter specifically, I guess they are by default doing caching, though it was not prompt caching, its KV Caching

[mango-panther]

@TomeHirata Could you also review this one when you get time

[TomeHirata]

Can you add tests?

[mango-panther]

I dont know what to do with cache=True as LLMLingua doesn't have that particular parameter specifically, I guess they are by default doing caching, though it was not prompt caching, its KV Caching

@TomeHirata Could you answer this one

[mango-panther]

@TomeHirata I have apply changed based on comments, updated docs based on other api references and also added tests.

[mango-panther]

@TomeHirata Can you review this again?