snc-library: Delete the failed pods before check for available one

Sometime pods goes to `ContainerStatusUnknown` state where it is not
able to send the status to kubelet and it stays there till manually
deleted and due to it our snc script fails. In this PR we are deleting
the pods which are in failed state (which is the same for
ContainerStatusUnknown one) and then checks the pods availablity.

```
+ sleep 256
+ all_pods_are_running_completed none
+ local ignoreNamespace=none
+ ./openshift-clients/linux/oc get pod --no-headers --all-namespaces '--field-selector=metadata.namespace!=none'
+ grep -v Running
+ grep -v Completed
openshift-kube-apiserver                           installer-11-crc                                         0/1   ContainerStatusUnknown   1                19m
+ exit=1
+ wait=512
+ count=10
+ '[' 10 -lt 10 ']'
+ echo 'Retry 10/10 exited 1, no more retries left.'
Retry 10/10 exited 1, no more retries left.
```

fixes: crc-org#920

snc-library.sh

@@ -241,8 +241,13 @@ function no_operators_degraded() {
     ${OC} get co -ojsonpath='{.items[*].status.conditions[?(@.type=="Degraded")].status}' | grep -v True
 }
 
+function retry_failed_pods() {
+    ${OC} delete pods --field-selector=status.phase=Failed -A
+}
+
 function all_pods_are_running_completed() {
     local ignoreNamespace=$1
+    retry_failed_pods
     ! ${OC} get pod --no-headers --all-namespaces --field-selector=metadata.namespace!="${ignoreNamespace}" | grep -v Running | grep -v Completed
 }