fix: GKE cluster update behavior

[krunalhinguu]

Issue

#1499 This PR resolves an issue where any change to a GKE cluster's node_pool in Terraform caused a full cluster replacement, which was disruptive and not necessary in most cases.

---

Problem

In the existing implementation of the rancher2_cluster resource for GKE clusters, the node_pool block was marked with ForceNew, which forced a full replacement of the cluster whenever any node pool change was made. This made it impossible to perform in-place updates like scaling or updating node pool properties without recreating the entire cluster.

---

Solution

• Removed ForceNew from the node_pool field in the Terraform schema.
• Added custom diff logic in the update function to properly detect and apply in-place changes to node_pool entries.

These changes allow users to modify existing GKE node pools without triggering full cluster recreation, improving usability and minimizing downtime.

---

Testing

Manual Testing

• Created a GKE cluster via Rancher using Terraform.
• Modified node pool fields (e.g., machine type, labels, node count).
• Verified that Terraform plan shows in-place updates, and apply works without cluster replacement.
• Confirmed updated configuration reflected correctly in Rancher UI and GCP Console.

Automated Testing

• No new automated tests added for this change.
• Existing tests for rancher2_cluster still pass, confirming backward compatibility.

---

QA Testing Considerations

• Test updating an existing GKE cluster's node pool (e.g., changing node count, version, labels).
• Confirm no full cluster replacement occurs.
• Validate that updates are reflected correctly in Rancher and on GKE.
• Test both new cluster creation and update flows to confirm schema compatibility.

---

Regressions Considerations

• Changes to node_pool update behavior could potentially cause issues if Rancher expects fields that are no longer being sent.
• Regressions are low to medium risk, but QA should test edge cases like:
  • Reducing node count to 0
  • Removing a node pool
  • Updating multiple properties at once

[matttrach]

This change is currently targeting Rancher v2.12.0 which coincides with v8.0.0 of this project. Please let me know if you need this to be backported to other versions.

[mjura]

LGTM

[matttrach]

Please sign your commits with a valid key so that we can merge.