dev(narugo): add more alignments on normalize

        pnormalize = PillowNormalize(mean=mean, std=std)

        tnormalize = Normalize(mean=mean, std=std)

        np.testing.assert_array_almost_equal(pnormalize(arr.numpy()), tnormalize(arr).numpy())

    @pytest.mark.parametrize(*tmatrix({

        'src_image': [

            'png_640.png',

            'png_640_m90.png',

        ],

    }))

    def test_maybe_normalize_invalid_image(self, src_image):

        image = Image.open(get_testfile(src_image))

        pnormalize = PillowNormalize(mean=0.5, std=0.5)

        with pytest.raises(TypeError):

            _ = pnormalize(image)

    @skipUnless(_TORCHVISION_AVAILABLE, 'Torchvision required.')

    @pytest.mark.parametrize(*tmatrix({

        'src_image': [

            'png_640.png',

            'png_640_m90.png',

        ],

        'mode': [

            'I', 'I;16'

        ],

        'mean': [

            (0.4850, 0.4560, 0.4060),

            (0.5, 0.5, 0.5),

            (0.0, 0.0, 0.0),

            (0.4850,),

            (0.5,),

            (0.0,),

        ],

        'std': [

            (0.2290, 0.2240, 0.2250),

            (0.5, 0.5, 0.5),

            (1.0, 1.0, 1.0),

            (0.2290,),

            (0.5,),

            (1.0,),

        ]

    }))

    def test_maybe_normalize_invalid_non_float(self, src_image, mode, mean, std):

        from torchvision.transforms import ToTensor

        image = Image.open(get_testfile(src_image))

        image = image.convert(mode)

        assert image.mode == mode

        arr = ToTensor()(image)

        pnormalize = PillowNormalize(mean=mean, std=std)

        with pytest.raises(TypeError):

            _ = pnormalize(arr.numpy())

    @skipUnless(_TORCHVISION_AVAILABLE, 'Torchvision required.')

    @pytest.mark.parametrize(*tmatrix({

        'src_image': [

            'png_640.png',

            'png_640_m90.png',

        ],

        'mode': [

            'F', '1', 'L', 'P',  # 1dim

        ],

        'mean': [

            (0.4850, 0.4560, 0.4060),

            (0.5, 0.5, 0.5),

            (0.0, 0.0, 0.0),

            (0.4850,),

            (0.5,),

            (0.0,),

        ],

        'std': [

            (0.2290, 0.2240, 0.2250),

            (0.5, 0.5, 0.5),

            (1.0, 1.0, 1.0),

            (0.2290,),

            (0.5,),

            (1.0,),

        ]

    }))

    def test_maybe_normalize_invalid_CHW(self, src_image, mode, mean, std):

        from torchvision.transforms import ToTensor

        image = Image.open(get_testfile(src_image))

        image = image.convert(mode)

        assert image.mode == mode

        arr = ToTensor()(image)

        assert arr.shape[0]

        arr = arr[0]

        pnormalize = PillowNormalize(mean=mean, std=std)

        with pytest.raises(ValueError):

            _ = pnormalize(arr.numpy())