fix(vm): avoid restartAwaitingChanges flicker for hotplug block devices

[eofff]

Description

When enableParavirtualization is enabled, VM block devices are handled through hotplug. Because of that, changes in blockDeviceRefs must not be reported by the VM change comparator as changes that require restart handling.

This PR stops reporting blockDeviceRefs add, remove, and reorder operations when paravirtualization is enabled in both the current and desired VM specs. Unit tests cover these hotplug block device scenarios and verify that they do not produce VM spec changes.

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

With paravirtualization enabled, attaching or detaching disks is a hotplug operation and should not affect the VM restart-awaiting-changes state.

Before this change, blockDeviceRefs updates were still detected as VM spec changes. As a result, restartAwaitingChanges could briefly appear or flicker while disks were being connected, disconnected, or reordered, even though these operations do not require a VM restart in the paravirtualized flow.

What is the expected result?

For a VM with enableParavirtualization: true in both the current and desired specs:

• Adding a block device does not trigger restartAwaitingChanges.
• Removing a block device does not trigger restartAwaitingChanges.
• Reordering block devices does not trigger restartAwaitingChanges.
• Hotplug disk operations do not make the VM look like it has pending restart-required changes.

For VMs without paravirtualization, the existing restart-required behavior for block device changes is preserved.

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.