!10271 Fix copy _batch_size of dataset

From: @luoyang42
Reviewed-by: @heleiwang,@liucunwei
Signed-off-by: @liucunwei

mindspore/dataset/engine/datasets.py

@@ -1359,6 +1359,9 @@ class Dataset:
     def input_indexs(self, value):
         self._input_indexs = value
 
+    def copy_batch_size(self, value):
+        self._batch_size = value
+
     def _init_tree_getters(self):
         """
         Get pipeline information.
@@ -1931,6 +1934,7 @@ class BatchDataset(Dataset):
         new_op.saved_output_types = self.saved_output_types
         new_op.saved_output_shapes = self.saved_output_shapes
         new_op.input_indexs = copy.deepcopy(self._input_indexs, memodict)
+        new_op.copy_batch_size(copy.deepcopy(self._batch_size, memodict))
         new_op.dataset_size = self.dataset_size
         new_op.pad = self.pad
         new_op.python_multiprocessing = copy.deepcopy(self.python_multiprocessing, memodict)