Feat: Added FaissDB to backend

[willingWill17]

🚀 PR Description: Add FAISS Vector Storage Backend,

📋 Summary

This PR introduces a new FAISS (Facebook AI Similarity Search) backend for the vector storage system. It enables high-performance similarity search with local file persistence, supporting multiple distance metrics and efficient in-memory indexing.

Usage

VECTOR_STORE_TYPE=faiss
FAISS_BASE_STORAGE_PATH=path/to/your/folder

---

✨ Features

• Core Backend Implementation

  • FaissBackend class implementing the VectorStore interface
  • Distance metrics supported: Cosine, Euclidean, Inner Product (IP)
  • Local file persistence with JSON-based storage
  • In-memory FAISS index via faiss-node
  • Automatic vector normalization for cosine distance

• Testing Coverage

  • Unit tests for connect, insert, search, update, delete, and list operations
  • Metric-specific tests for Euclidean and IP scoring
  • Error handling tests for invalid operations and connection issues
  • Filter support tests: range queries, any filters
  • Mock implementations for file system and FAISS interactions

• Configuration Integration

  • Environment variable support: FAISS_BASE_STORAGE_PATH
  • Schema validation with Zod
  • Factory integration for backend selection
  • Default configuration with sensible defaults

---

⚙️ Key Implementation Details

• Distance Metrics

  • Cosine → IndexFlatIP + normalization
  • Euclidean → IndexFlatL2
  • Inner Product → IndexFlatIP

• Data Persistence

  • JSON storage per collection
  • Atomic file operations with error handling
  • Lazy loading on connection
  • Automatic directory creation for persistence path

• Search Capabilities

  • Configurable similarity search result limits
  • Distance-based scoring per metric
  • Metadata filters (including range operators: gte, lte, gt, lt)

---

✅ Testing Highlights

• Constructor validation for all distance metrics
• Robust connection lifecycle tests
• CRUD operation coverage
• Search with score validation
• Advanced filter queries
• File persistence using mocked FS

---

🔗 Related Issues

• Extended issue [FEATURE] Implement FAISS Vector Store Backend #216