Whenever the audit webhook is enabled, we should increase the time it takes to terminate the kubernetes apiserver and modify the audit-webhook-initial-backoff (default 10 seconds) to 5 seconds. 

When the audit webhook is enabled and kube-apiserver is given a SIG Term signal, it will hold 70 seconds before terminating (determined by shutdown-delay-duration), then the apiserver will start shutting down. The audit webhook will attempt 10 times and retry after 10 seconds. So I'm suggestion we modify audit-webhook-initial-backoff to 5 second so it only needs an extra 50 seconds and increase the termination grace period to 130 seconds. This will give us a 5-10 second buffer for the audit webhook to terminate gracefully.

    1. Create a hypershift cluster with https://github.com/openshift/hypershift/pull/5475 and https://github.com/openshift/hypershift/pull/5491     2. enable auditting and a webhook on the cluster (you'd have to figure out this part yourself)
    3. Time the deletion of a kube-apiserver pod
    

I tested this myself and I appear to be getting between 120-126 seconds. Justifying why I think 130 would be a good number for termination grace period.

# joseph.goergen@stgiks-dal10-carrier0-worker-1002:~$ time kubectl delete pod -n master-cv5k78u20ksllb9rsk1g kube-apiserver-5d7fcf5b9f-52cm8
pod "kube-apiserver-5d7fcf5b9f-52cm8" deletedreal    2m2.267s
user    0m0.361s
sys    0m0.092s
# joseph.goergen@stgiks-dal10-carrier0-worker-1002:~$ time kubectl delete pod -n master-cv5k78u20ksllb9rsk1g kube-apiserver-5d7fcf5b9f-rtk5x
pod "kube-apiserver-5d7fcf5b9f-rtk5x" deletedreal    2m6.099s
user    0m0.327s
sys    0m0.082s
# joseph.goergen@stgiks-dal10-carrier0-worker-1002:~$ time kubectl delete pod -n master-cv5k78u20ksllb9rsk1g kube-apiserver-5d7fcf5b9f-h4smt
pod "kube-apiserver-5d7fcf5b9f-h4smt" deletedreal    1m59.644s
user    0m0.244s
sys    0m0.137s