reranking client fix

smaller reranking model and new ingestion pipeline to test later

reranking made suitable for jinai model

reranker model name changed

reranker import fix

Dockerfile model load and import fix

reranking model integrated

upsert logic added to the ingestion

sparse tuple fix

async client and sync for qdrant seperated for ingestion

model name fix

hybrid search activated with fastembed, ingestion made sync

chat history ready for frontend integration, me endpoint

hybrid fix

hybrid turned off

hf github sync

fix

hf initialized