[backport v2.12.6] [BUG][UI]Pod cannot revert Node Scheduling from "Run pods on specific nodes" back to "Run pods on any available node" via UI

[rancher-ui-project-bot]

This is a backport issue for #16092, automatically created via GitHub Actions workflow initiated by @richard-cox

Original issue body:

Rancher Server Setup

• Rancher version: 2.11/2.12/2.13

Describe the bug
Our customer Coca-Cola reported an issue with the Node Scheduling UI in Rancher v2.11.6.

When editing a workload and changing Node Scheduling from “Run pods on any available node” to “Run pods on specific nodes”, it is not possible to revert the setting back to “Run pods on any available node” using the UI. The only way to restore the original behavior is to edit the workload YAML and manually remove the nodeName field.

This effectively makes the “Run pods on any available node” option non-functional once “Run pods on specific nodes” has been selected for that workload.

To Reproduce

1. Log in to Rancher
2. Navigate to a cluster and open Workloads.
3. Create or edit a Deployment/Workload.
4. In the Node Scheduling section, select “Run pods on any available node” and save the workload.
5. Edit the same workload again and change Node Scheduling to “Run pods on specific nodes”, select a node, and save.
6. Edit the workload one more time and try to change Node Scheduling back to “Run pods on any available node” via the UI.

Expected Result
When the user selects “Run pods on any available node” in the Node Scheduling UI and saves:
The nodeName field should be cleared from the workload’s Pod template.
Pods should be schedulable on any available node according to normal scheduler behavior.

[richard-cox]

@smallteeths Once #16108 merges can you backport to the release-2.12 branch?

[smallteeths]

@smallteeths Once #16108 merges can you backport to the release-2.12 branch?

Sure, once #16108 is merged I’ll backport it to the release-2.12 branch.