- Wait for kubelet & kube-proxy to be ready on Windows node within 10s ([#85228](https://github.com/kubernetes/kubernetes/pull/85228), [@YangLu1031](https://github.com/YangLu1031)) [SIG Cluster Lifecycle]

  - `failurePolicy` default changed from `Ignore` to `Fail` for v1
  - `matchPolicy` default changed from `Exact` to `Equivalent` for v1
  - `timeout` default changed from `30s` to `10s` for v1
  - `sideEffects` default value is removed, and the field made required, and only `None` and `NoneOnDryRun` are permitted for v1

    * To avoid that negative effect only use per pod metrics from pods that are:
    * - ready (so metrics about them should be valid), or
    * - unready but creation and last readiness change timestamps are apart more than 10s (pods that have formerly been ready and so metrics are in at least some cases (pod becoming unready because of overload) very useful).