Feat/34 llm reranker

[hangolehai]

Summary

Adds an LLM re-ranking step to the hybrid retriever to score and sort document chunks using OpenAI.

Issue

Closes #34

Changes

• Created LLMReranker in rag/retriever/llm_reranker.py
• Updated HybridRetriever to optionally accept the reranker and use it to sort final_results.
• Added unit tests for the reranker using mock OpenAI responses.

Testing

• Unit tests pass (make test-unit)
• Integration tests pass (make test-integration)
• Linter passes (make lint)
• Type checker passes (make typecheck)
• New/updated tests cover the changes

Screenshots / Demo

Notes for Reviewers

Note: There are pre-existing test failures (53 in make test-unit) that I observed before making my changes. My code passed its own unit tests and did not introduce any new failures.