feat(cluster): enable GKE managed ML Diagnostics for GKE version >= 1.35.0-gke.3065000

[rapatchi]

Description

This change introduces automated management and lifecycle optimization for GKE Managed Machine Learning Diagnostics on cluster creation.

Specifically:

1. Standardizes PEP 440 version comparison via is_gke_version_at_least() with a central minimum requirement constant (MANAGED_MLDIAGNOSTICS_MIN_GKE_VERSION = "1.35.0-gke.3065000").
2. Refactors run_gke_cluster_create_command() to dynamically append --enable-managed-mldiagnostics when --managed-mldiagnostics is requested and the planned GKE control plane version meets or exceeds the minimum threshold.
3. Bypasses legacy user-space prerequisite installations (install_mldiagnostics_prerequisites()) when creating or updating clusters on modern GKE versions.
4. Refines --dry-run simulation in get_gke_server_config() and get_gke_node_pool_version() to correctly evaluate user-supplied GKE versions (--gke-version=X) or default smoothly to "0", ensuring accurate simulation without breaking Golden Recipe verification.

Issue

Correcting the behaviour of: --managed-mldiagnosticsflag

Testing

Have you performed any manual testing on your change? Yes

xpk cluster create with >= 1.35.0-gke.3065000:

(xpk_local_venv) rapatchi@rapatchi2:~/xpk_fork/xpk$ xpk cluster create --cluster=maxtest-cluster --tpu-type=v5litepod-8 --project=rapatchiconsumer --zone=us-central1-a --num-nodes=2 --managed-mldiagnostics --spot
[XPK] Starting xpk v0.1.dev902+g7c8ca0ae1
...
[XPK] Task: `GKE Cluster Create` is implemented by `gcloud beta container clusters create maxtest-cluster --project=rapatchiconsumer --region=us-central1 --node-locations=us-central1-a --cluster-version=1.36.0-gke.2459000 --machine-type=e2-standard-16 --enable-autoscaling --total-min-nodes 1 --total-max-nodes 1000 --num-nodes 6 --autoscaling-profile=optimize-utilization --labels=gke_product_type=xpk --release-channel=rapid --enable-ip-alias --enable-dataplane-v2 --enable-multi-networking --enable-dns-access --location-policy=BALANCED --scopes=storage-full,gke-default --enable-managed-mldiagnostics`, streaming output live.
...
[XPK] Task: `Updating Controller Manager resources` terminated with code `0`
[XPK] GKE commands done! Resources are created.
[XPK] See your GKE Cluster here: https://console.cloud.google.com/kubernetes/clusters/details/us-central1/maxtest-cluster/details?project=rapatchiconsumer
[XPK] Exiting XPK cleanly

xpk cluster create with < 1.35.0-gke.3065000

(xpk_local_venv) rapatchi@rapatchi2:~/xpk_fork/xpk$ xpk cluster create --cluster=maxtest-cluster --tpu-type=v5litepod-8 --project=rapatchiconsumer --zone=us-central1-a --num-nodes=2 --managed-mldiagnostics --spot --gke-version=1.34.8-gke.1000000
[XPK] Starting xpk v0.1.dev902+g7c8ca0ae1
...
[XPK] Task: `GKE Cluster Create` is implemented by the following command not running since it is a dry run.
gcloud beta container clusters create maxtest-cluster --project=rapatchiconsumer --region=us-central1 --node-locations=us-central1-a --cluster-version=1.34.8-gke.1000000 --machine-type=e2-standard-16 --enable-autoscaling --total-min-nodes 1 --total-max-nodes 1000 --num-nodes 6 --autoscaling-profile=optimize-utilization --labels=gke_product_type=xpk --release-channel=regular --enable-ip-alias --enable-dataplane-v2 --enable-multi-networking --no-enable-autoupgrade --enable-dns-access --location-policy=BALANCED --scopes=storage-full,gke-default
...
[XPK] Task: `Applying cert-manager 1.13.0 manifest...` is implemented by `kubectl apply -f [https://github.com/cert-manager/cert-manager/releases/download/v1.13.0/cert-manager.yaml](https://www.google.com/url?sa=D&q=https%3A%2F%2Fgithub.com%2Fcert-manager%2Fcert-manager%2Freleases%2Fdownload%2Fv1.13.0%2Fcert-manager.yaml)`, streaming output live.
[XPK] Task: `Create gke-mldiagnostics namespace...` is implemented by `kubectl create namespace gke-mldiagnostics`
[XPK] Task: `Install /tmp/mldiagnostics-injection-webhook-v0.5.0.yaml...` is implemented by `kubectl apply -f /tmp/mldiagnostics-injection-webhook-v0.5.0.yaml -n gke-mldiagnostics`, streaming output live.

Have you verified use cases affected by goldens? Yes

[scaliby]

Thanks for your changes! Two minor fixes to go.

[scaliby]

thanks for addressing my feedback! LGTM!