chore(fleet): set Kubernetes 1.34 creation defaults only

[plural-copilot]

Summary

• Default-only change: normalizes new Fleet cluster creation/setup defaults to Kubernetes 1.34; it does not upgrade existing fleets or alter instantiated state under bootstrap/fleets/.
• Updates template/fleet/dev.yaml and template/fleet/prod.yaml so both kubernetesVersion and nextKubernetesVersion default to 1.34.
• Keeps GKE's existing two-field staged blue/green semantics intact: both kubernetes_version and next_kubernetes_version now default to 1.34; release_channel = "UNSPECIFIED" and provider/module constraints are unchanged.
• Updates EKS, LKE, and legacy GKE-module Terraform defaults to 1.34.
• Repairs the LKE contract by renaming kubernetes_vsn to kubernetes_version and updating its single internal reference. Shared templates already forward kubernetes_version, so no template wiring or compatibility alias was required.
• Updates legacy template/clusters.yaml creation defaults from 1.28 to 1.34 without changing its structure.

Changed paths

• template/fleet/dev.yaml
• template/fleet/prod.yaml
• template/clusters.yaml
• terraform/gke-cluster/variables.tf
• terraform/eks-cluster/variables.tf
• terraform/lke-cluster/variables.tf
• terraform/lke-cluster/lke.tf
• terraform/modules/gke/variables.tf

Validation

Passed:

timeout 5m docker run --rm -v "$PWD:/workspace:ro" -w /workspace ruby:3.3-alpine ruby -ryaml -e '...' template/fleet/dev.yaml template/fleet/prod.yaml template/clusters.yaml
# YAML template defaults parsed and matched expected 1.34 values

test -z "$(grep -RIn --exclude-dir=.git 'kubernetes_vsn' .)"
grep -Fq 'variable "kubernetes_version"' terraform/lke-cluster/variables.tf
grep -Fq 'k8s_version = var.kubernetes_version' terraform/lke-cluster/lke.tf
grep -Fq 'release_channel = "UNSPECIFIED"' terraform/gke-cluster/cluster.tf
git diff --check
# LKE contract, GKE release channel, changed-path, bootstrap exclusion, and diff checks passed

timeout 10m docker run --rm --entrypoint /bin/sh -v "$PWD:/workspace" -w /workspace hashicorp/terraform:1.8.2 -ec '... terraform/gke-cluster ... init -backend=false ...; validate ...'
# PASS (existing deprecated kubeconfig warning only)

timeout 10m docker run --rm --entrypoint /bin/sh -v "$PWD:/workspace" -w /workspace hashicorp/terraform:1.8.2 -ec '... terraform/lke-cluster ... init -backend=false ...; validate ...'
# PASS (existing deprecated kubeconfig warning only)

Caveats:

• terraform fmt -check terraform/gke-cluster terraform/eks-cluster terraform/lke-cluster terraform/modules/gke reports pre-existing formatting differences throughout those directories. Formatting only the touched files also rewrote unrelated baseline formatting, so that output was reverted to keep this PR default-only and reviewable; changed Terraform lines retain the existing repository formatting.
• terraform validate for terraform/eks-cluster is blocked by existing compatibility issues when dependency resolution selects AWS provider 6.62.0 (legacy module arguments/blocks unsupported).
• terraform/modules/gke cannot complete init because its existing merged Google provider constraints are mutually incompatible. No provider/module constraints were changed.

[plural-copilot]

This PR was generated by the codex Plural Agent Runtime. Here's some useful information you might want to know to evaluate the ai's perfomance:

Name	Details
💬 Prompt	Create one reviewable PR to normalize Fleet cluster creation/setup defaults to Kubernetes 1.34....
🔗 Run history	View run history