Description of problem:
Nodes couldn't recover when missing worker role in the custom mcp, all of the configuration missed in the node, the kubelet and crio services couldn't start.

Version-Release number of selected component (if applicable):
OCP 4.14

How reproducible:
Steps to Reproduce:

1. Create a custom MCP without worker role
$ cat mc.yaml
apiVersion: machineconfiguration.openshift.io/v1
kind: MachineConfig
metadata:
labels:
machineconfiguration.openshift.io/role: worker-t
generation: 3
name: 80-user-kernal
spec: {}

$ cat mcp.yaml
apiVersion: machineconfiguration.openshift.io/v1
kind: MachineConfigPool
metadata:
name: worker-t
spec:
configuration:
machineConfigSelector:
matchLabels:
machineconfiguration.openshift.io/role: worker-t
nodeSelector:
matchLabels:
node-role.kubernetes.io/worker-t: ""

$ oc create -f mc.yaml
$ oc create -f mcp.yaml

2. Add label worker-t to worker03

$ oc get no
NAME STATUS ROLES AGE VERSION
master01.ocp4.danliu.com Ready master 454d v1.27.13+e709aa5
master02.ocp4.danliu.com Ready master 453d v1.27.13+e709aa5
master03.ocp4.danliu.com Ready master 453d v1.27.13+e709aa5
worker01.ocp4.danliu.com Ready worker 453d v1.27.13+e709aa5
worker02.ocp4.danliu.com Ready worker 51d v1.27.13+e709aa5
worker03.ocp4.danliu.com Ready worker,worker-t 69d v1.27.13+e709aa5

$ oc label nodes worker03.ocp4.danliu.com node-role.kubernetes.io/worker-t=
node/worker03.ocp4.danliu.com labeled

Actual results:
worker03 run into NotReady status, kubelet and crio couldn't startup.

Expected results:
Prevent to sync up the mc when missing worker role

Additional info:
In the previous version (4.13 & 4.12), the task stuck with below error:

Marking Unreconcilable due to: can't reconcile config rendered-worker-8f464eb07d2e2d2fbdb84ab2204fea65 with rendered-worker-t-5b6179e2fb4fedb853c900504edad9ce: ignition passwd user section contains unsupported changes: user core may not be deleted