fix(vm): sync unschedulable KVVM on vmclass placement change

[danilrwx]

Description

Fix synchronization of internal KubeVirt VM when the VM is in Unschedulable state and placement-related settings are changed via VMClass.

The workaround path in SyncKvvmHandler now treats these change paths as placement policy updates:

• virtualMachineClassName
• VirtualMachineClass:spec.nodeSelector
• VirtualMachineClass:spec.tolerations

This allows controller to apply KVVM update and recycle launcher pod for unschedulable cases.

Also added unit tests for placement-path detection in isPlacementPolicyChanged.

Why do we need it, and what problem does it solve?

When VM pod creation fails and KVVM stays Unschedulable, changing VMClass (for example, discovery -> generic) did not trigger full KVVM resync for placement-dependent fields.
As a result, domain.cpu / tolerations could stay stale and VM remained stuck.

This change makes unschedulable recovery react to VMClass-driven placement changes.

What is the expected result?

1. Put VM into Unschedulable state.
2. Change VMClass name or update VMClass spec.nodeSelector / spec.tolerations.
3. Verify KVVM is synchronized and VM can leave Unschedulable when constraints become schedulable.

Checklist

• The code is covered by unit tests.
• e2e tests passed.
• Documentation updated according to the changes.
• Changes were tested in the Kubernetes cluster manually.

Changelog entries

section: core
type: fix
summary: "Fix KVVM resync for unschedulable VMs when VMClass placement settings are changed."