core[patch]: Convert SimSIMD back to NumPy (langchain-ai#19473)

This patch fixes the langchain-ai#18022 issue, converting the SimSIMD internal
zero-copy outputs to NumPy.

I've also noticed, that oftentimes `dtype=np.float32` conversion is used
before passing to SimSIMD. Which numeric types do LangChain users
generally care about? We support `float64`, `float32`, `float16`, and
`int8` for cosine distances and `float16` seems reasonable for
practically any kind of embeddings and any modern piece of hardware, so
we can change that part as well 🤗

libs/community/langchain_community/utils/math.py

@@ -29,7 +29,7 @@ def cosine_similarity(X: Matrix, Y: Matrix) -> np.ndarray:
         Z = 1 - simd.cdist(X, Y, metric="cosine")
         if isinstance(Z, float):
             return np.array([Z])
-        return Z
+        return np.array(Z)
     except ImportError:
         logger.info(
             "Unable to import simsimd, defaulting to NumPy implementation. If you want "

libs/partners/elasticsearch/langchain_elasticsearch/_utilities.py

@@ -79,7 +79,7 @@ def cosine_similarity(X: Matrix, Y: Matrix) -> np.ndarray:
         Z = 1 - simd.cdist(X, Y, metric="cosine")
         if isinstance(Z, float):
             return np.array([Z])
-        return Z
+        return np.array(Z)
     except ImportError:
         X_norm = np.linalg.norm(X, axis=1)
         Y_norm = np.linalg.norm(Y, axis=1)

libs/partners/mongodb/langchain_mongodb/utils.py

@@ -38,7 +38,7 @@ def cosine_similarity(X: Matrix, Y: Matrix) -> np.ndarray:
         Z = 1 - simd.cdist(X, Y, metric="cosine")
         if isinstance(Z, float):
             return np.array([Z])
-        return Z
+        return np.array(Z)
     except ImportError:
         logger.info(
             "Unable to import simsimd, defaulting to NumPy implementation. If you want "

libs/partners/pinecone/langchain_pinecone/_utilities.py

@@ -69,7 +69,7 @@ def cosine_similarity(X: Matrix, Y: Matrix) -> np.ndarray:
         Z = 1 - simd.cdist(X, Y, metric="cosine")
         if isinstance(Z, float):
             return np.array([Z])
-        return Z
+        return np.array(Z)
     except ImportError:
         X_norm = np.linalg.norm(X, axis=1)
         Y_norm = np.linalg.norm(Y, axis=1)