For historical context, see the discussion in https://issues.redhat.com/browse/OCPBUGS-43631.

Especially on single node clusters, we're a lot of cases where kube-apiserver isn't coming up within the expected alloted time. (3m 15s)
Here are some examples from the recent SNO presubmits:

• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_cluster-authentication-operator/740/pull-ci-openshift-cluster-authentication-operator-master-e2e-aws-single-node/1942860654289358848
• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/29973/pull-ci-openshift-origin-main-e2e-aws-ovn-single-node/1942771670083178496
• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/29955/pull-ci-openshift-origin-main-e2e-aws-ovn-single-node/1942594136792109056
• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/29967/pull-ci-openshift-origin-main-e2e-aws-ovn-single-node-serial/1942387065366777856
• https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/29965/pull-ci-openshift-origin-main-e2e-aws-ovn-single-node/1942293816438427648