*   New endpoints in `tf.image` namespace: `tf.image.extract_image_patches`
    *   New endpoints in `tf.debugging` namespace:
        `tf.debugging.check_numerics`, `tf.debugging.is_finite`,
        `tf.debugging.is_inf`, `tf.debugging.is_nan`.
    *   New endpoints in `tf.dtypes` namespace: `tf.dtypes.as_string`.
    *   New endpoints in `tf.io` namespace: `tf.io.decode_base64`,

- Restore the ability to `kubectl apply --prune` without --namespace flag.  Since 1.17, `kubectl apply --prune` only prunes resources in the default namespace (or from kubeconfig) or explicitly specified in command line flag.  But this is s breaking change from kubectl 1.16, which can prune resources in all namespace in config file.  This patch restores the kubectl 1.16 behaviour. ([#89551](https://github.com/kubernetes/kubernetes/pull/89551),...