Action required for custom scheduler plugin developers:
  Plugins **have to** implement a QueueingHint for Pod/Update event if the rejection from them could be resolved by updating unscheduled Pods themselves.
  Example: suppose you develop a custom plugin that denies Pods that have a `schedulable=false` label. 

- Added consistent vanity import to files and provided tooling for verifying and updating them. ([#120642](https://github.com/kubernetes/kubernetes/pull/120642), [@jcchavezs](https://github.com/jcchavezs))

- The `matchLabelKeys/mismatchLabelKeys` feature is introduced to the hard/soft `PodAffinity/PodAntiAffinity`. ([#116065](https://github.com/kubernetes/kubernetes/pull/116065), [@sanposhiho](https://github.com/sanposhiho))
- When updating a CRD, per-expression cost limit check are now skipped for `x-kubernetes-validations` rules of versions that are not mutated. ([#121460](https://github.com/kubernetes/kubernetes/pull/121460), [@jiahuif](https://github.com/jiahuif))

- Kubeadm: improved retries when updating node information, in case `kube-apiserver` is temporarily unavailable ([#114176](https://github.com/kubernetes/kubernetes/pull/114176), [@QuantumEnergyE](https://github.com/QuantumEnergyE))

- Fix problem in updating VolumeAttached in node status ([#112305](https://github.com/kubernetes/kubernetes/pull/112305), [@xing-yang](https://github.com/xing-yang)) [SIG Apps]