refactor: make cosine similarity threshold a required config parameter

• Remove default threshold from env var
• Add validation for missing threshold
• Move default to lightrag.py config init
• Update all vector DB implementations
• Improve threshold validation consistency

lightrag/kg/nano_vector_db_impl.py

@@ -73,16 +73,17 @@ from lightrag.base import (
 
 @dataclass
 class NanoVectorDBStorage(BaseVectorStorage):
-    cosine_better_than_threshold: float = float(os.getenv("COSINE_THRESHOLD", "0.2"))
+    cosine_better_than_threshold: float = None
 
     def __post_init__(self):
         # Initialize lock only for file operations
         self._save_lock = asyncio.Lock()
         # Use global config value if specified, otherwise use default
         config = self.global_config.get("vector_db_storage_cls_kwargs", {})
-        self.cosine_better_than_threshold = config.get(
-            "cosine_better_than_threshold", self.cosine_better_than_threshold
-        )
+        cosine_threshold = config.get("cosine_better_than_threshold")
+        if cosine_threshold is None:
+            raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
+        self.cosine_better_than_threshold = cosine_threshold
 
         self._client_file_name = os.path.join(
             self.global_config["working_dir"], f"vdb_{self.namespace}.json"