Kubernetes broken for CI

[mhutchinson]

K8s CI steps are currently failing. This is because I deleted the ancient k8s cluster due to security flags. I've recreate with a new instance, but our k8s configs do not work with new versions of k8s. Current error is in "Step 7: apply_k8s_cfgs_for_spanner_dryrun":

Running: gcloud container clusters get-credentials --project="trillian-opensource-ci" --zone="us-central1-a" "trillian-opensource-ci"
--
  | Fetching cluster endpoint and auth data.
  | kubeconfig entry generated for trillian-opensource-ci.
  | Running: kubectl apply --dry-run=server -f=envsubst-spanner/etcd-cluster.yaml -f=envsubst-spanner/trillian-ci-spanner.yaml -f=envsubst-spanner/trillian-log-deployment.yaml -f=envsubst-spanner/trillian-log-service.yaml -f=envsubst-spanner/trillian-log-signer-deployment.yaml -f=envsubst-spanner/trillian-log-signer-service.yaml --prune --all --prune-allowlist=core/v1/ConfigMap
  | secret/trillian-secrets created (server dry run)
  | configmap/deploy-config created (server dry run)
  | deployment.apps/trillian-logserver-deployment created (server dry run)
  | service/trillian-log-service created (server dry run)
  | deployment.apps/trillian-logsigner-deployment created (server dry run)
  | service/trillian-log-signer created (server dry run)
  | error: resource mapping not found for name: "trillian-etcd-cluster" namespace: "" from "envsubst-spanner/etcd-cluster.yaml": no matches for kind "EtcdCluster" in version "etcd.database.coreos.com/v1beta2"
  | ensure CRDs are installed first

I will take a quick pass at upgrading these configs so they should work with new versions, but if I can't get something working quickly then we will need to disable the k8s stuff so we can unblock core Go dependency updates such as #3835.