Start using the vectorized EncodePng Op (#6344)

## Motivation for features / changes
The vectorization unblocks running tf.summary.image under DTensor.

Vectorization of EncodePNG is added in
tensorflow/tensorflow@7c42227

## Technical description of changes
Applied a forward compatibility wrapper for 14 days since the EncodePNG change landed.

## Detailed steps to verify changes work correctly (as executed by you)
Verified by both branches work without DTensor, and the new (vectorized)
branch works under DTensor.
Note that I manually applied the internal patch to the OSS format. 

## Verification: 
cl/488391097

tensorboard/plugins/image/summary_test.py

@@ -237,6 +237,25 @@ def test_floating_point_data(self):
         # range [0, 255] with 229 = 0.9 * 255, truncated.
         self.assertAllEqual([0, 0, 229, 255, 255], list(decoded.flat))
 
+    def test_vector_data(self):
+        data = np.array(
+            [
+                [-0.01, 0.0, 0.9, 1.0, 1.1],
+                [-0.01, 0.0, 1.0, 0.9, 1.1],
+            ]
+        ).reshape((2, -1, 1, 1))
+        pb = self.image("mona_lisa", data)
+
+        encoded0 = pb.value[0].tensor.string_val[2]  # skip width, height
+        decoded0 = tf.image.decode_png(encoded0).numpy()
+        # Float values outside [0, 1) are truncated, and everything is scaled to the
+        # range [0, 255] with 229 = 0.9 * 255, truncated.
+        self.assertAllEqual([0, 0, 229, 255, 255], list(decoded0.flat))
+
+        encoded1 = pb.value[0].tensor.string_val[3]  # skip width, height
+        decoded1 = tf.image.decode_png(encoded1).numpy()
+        self.assertAllEqual([0, 0, 255, 229, 255], list(decoded1.flat))
+
 
 if __name__ == "__main__":
     tf.test.main()

tensorboard/plugins/image/summary_v2.py

@@ -113,18 +113,24 @@ def lazy_tensor():
             tf.debugging.assert_non_negative(max_outputs)
             images = tf.image.convert_image_dtype(data, tf.uint8, saturate=True)
             limited_images = images[:max_outputs]
-            encoded_images = tf.map_fn(
-                tf.image.encode_png,
-                limited_images,
-                dtype=tf.string,
-                name="encode_each_image",
-            )
-            # Workaround for map_fn returning float dtype for an empty elems input.
-            encoded_images = tf.cond(
-                tf.shape(input=encoded_images)[0] > 0,
-                lambda: encoded_images,
-                lambda: tf.constant([], tf.string),
-            )
+            if tf.compat.forward_compatible(2023, 5, 1):
+                encoded_images = tf.image.encode_png(limited_images)
+            else:
+                # TODO(b/276803093): The kernel was updated around 2023/04/15.
+                # After 90 days (2023/07/15), please remove the False branch.
+                encoded_images = tf.map_fn(
+                    tf.image.encode_png,
+                    limited_images,
+                    dtype=tf.string,
+                    name="encode_each_image",
+                )
+                # Workaround for map_fn returning float dtype for an empty
+                # elems input.
+                encoded_images = tf.cond(
+                    tf.shape(input=encoded_images)[0] > 0,
+                    lambda: encoded_images,
+                    lambda: tf.constant([], tf.string),
+                )
             image_shape = tf.shape(input=images)
             dimensions = tf.stack(
                 [