increase quality of hybrid search using LLM-based reranking

[bard]

General documentation about reranking:

• https://www.sbert.net/examples/applications/retrieve_rerank/README.html
• https://medium.com/mlearning-ai/search-rank-and-recommendations-35cc717772cb

Note: re-ranking might introduce latency at a latency-critical point and therefore be a no-go.