enable type conversion in ThreadwiseGenericTensorSliceCopy_v2r1 and BlockwiseGenericTensorSliceCopy_v2