fix: convert NumPy arrays and other vector types in insert_many (Fixes #1002)

insert_many() in executor.py constructs _BatchObject directly without
converting the vector, causing NumPy arrays, torch tensors, and other
supported vector types to be silently discarded. The fix applies
_get_vector_v4() conversion (same as BatchObject.__init__) before
storing the vector in _BatchObject.

Also handles named vectors (dict[str, vector]) by converting each value.

Signed-off-by: rtmalikian <rtmalikian@gmail.com>

Author: rtmalikian

weaviate/collections/data/executor.py

@@ -177,7 +177,13 @@ def insert_many(
             (
                 _BatchObject(
                     collection=self.name,
-                    vector=obj.vector,
+                    vector=(
+                        {key: _get_vector_v4(val) for key, val in obj.vector.items()}
+                        if isinstance(obj.vector, dict)
+                        else _get_vector_v4(obj.vector)
+                        if obj.vector is not None
+                        else None
+                    ),
                     uuid=str(obj.uuid if obj.uuid is not None else uuid_package.uuid4()),
                     properties=cast(dict, obj.properties),
                     tenant=self._tenant,