[Testing] improve and enable the skipped testing cases

[Xunzhuo]

1. BERT Similarity Model Tests (8 tests)

Root Cause: Failed to initialize BERT similarity model (sentence-transformers/all-MiniLM-L6-v2)

• TestInitModel/InitWithSpecificModel - Model initialization with specific model
• TestTokenization - Tokenization functionality tests
• TestEmbeddings - Embedding generation tests
• TestSimilarity - Similarity calculation tests
• TestFindMostSimilar - Finding most similar items tests
• TestUtilityFunctions/IsModelInitialized - Model initialization status check
• TestErrorHandling/EmptyStringHandling - Empty string error handling
• TestConcurrency - Concurrent operations tests

2. Milvus Vector Database Tests (5 tests)

Root Cause: MILVUS_URI environment variable not set

• TestHybridCacheBasicOperations - Basic hybrid cache operations
• TestHybridCachePendingRequest - Pending request handling in hybrid cache
• TestHybridCacheEviction - Cache eviction logic
• TestHybridCacheLocalCacheHit - Local cache hit scenarios
• TestHybridVsMilvusSmoke - Smoke test comparing hybrid vs Milvus (also skipped in short mode)

3. Model Directory Tests (3 tests)

Root Cause: Models directory does not exist at ../../../../../models

• TestAutoDiscoverModels_RealModels - Auto-discovery of available models
• TestAutoInitializeUnifiedClassifier - Unified classifier initialization
• TestUnifiedClassifier_Integration - Integration tests for unified classifier

4. LoRA Model Tests (1 test)

Root Cause: Candle BERT token classifier not available at ../models/lora_pii_detector_bert-base-uncased_model

• TestCandleBertTokensWithLabels - Token classification with labels using LoRA model

5. Embedding Model Tests (4 tests)

Root Cause: Failed to initialize embedding models (Gemma model config not found at ../models/embeddinggemma-300m/config.json)

• TestGetEmbeddingSmart - Smart embedding retrieval
• TestGetEmbeddingWithDim - Embedding with specific dimensions
• TestEmbeddingConsistency - Embedding consistency validation
• TestEmbeddingPriorityRouting - Priority-based routing with embeddings

[yehudit1987]

/assign

[yehudit1987]

Hi @Xunzhuo, would it be okay if I open multiple PRs to handle the different sub-tasks of this task?
It’ll make the review process much easier for me, and I believe it’ll help the reviewer as well.