Use vector store for embedding item-to-item

[zhangzhenghao]

Summary

• write matrix factorization item embeddings into the vector store after collaborative filtering training
• store only matrix factorization user embeddings in blob storage, with backward-compatible worker loading for old blobs
• generate collaborative filtering recommendations via vector search instead of local item HNSW index
• use vector-store-backed query paths for embedding item-to-item recommendations

Tests

• go test ./logics ./master ./worker ./server -count=1

[codecov]

Codecov Report

❌ Patch coverage is 54.12844% with 150 lines in your changes missing coverage. Please review.
✅ Project coverage is 73.09%. Comparing base (85f731d) to head (a22ccab).
⚠️ Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
master/tasks.go	32.05%	47 Missing and 6 partials ⚠️
logics/item_to_item.go	67.96%	19 Missing and 14 partials ⚠️
worker/worker.go	2.94%	33 Missing ⚠️
logics/recommend.go	74.07%	7 Missing and 7 partials ⚠️
server/rest.go	61.76%	8 Missing and 5 partials ⚠️
logics/cf.go	81.81%	1 Missing and 1 partial ⚠️
worker/pipeline.go	66.66%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1306      +/-   ##
==========================================
- Coverage   73.62%   73.09%   -0.53%     
==========================================
  Files          93       93              
  Lines       17745    18102     +357     
==========================================
+ Hits        13064    13231     +167     
- Misses       3292     3440     +148     
- Partials     1389     1431      +42     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.