fix(core): run firmware upgrade migration only for running vmi

[danilrwx]

Description

Improve workload-updater migration behavior for firmware updates and increase unit-test coverage for handler decision branches.

Functional change

FirmwareHandler now handles KVVMI read outcomes explicitly:

• if Get(KVVMI) returns NotFound, processing continues;
• if Get(KVVMI) returns any other error, handler returns error;
• if KVVMI exists but status.phase != Running, migration is not triggered.

Test changes

Extended branch coverage for:

• pkg/controller/workload-updater/internal/handler/firmware.go
• pkg/controller/workload-updater/internal/handler/nodeplacement.go

Coverage after changes (go test ./pkg/controller/workload-updater/internal/handler/... -coverprofile=/tmp/wu-handler.cover):

• firmware.go — 100% per function (Handle, needUpdate, isVirtControllerUpToDate, getVirtLauncherImage, Name)
• nodeplacement.go — Handle 94.1%, Name 100%, genNodePlacementSum 90.0%
• package total handler — 90.5%

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

Previously, firmware updater could create firmware-update-* VMOP while VMI was not running (Pending/Succeeded). This produced unnecessary failed migration VMOPs and misleading operation status.

Also, KVVMI absence should not block processing when all other preconditions are satisfied.

Additional tests reduce regression risk in workload-updater branching logic.

What is the expected result?

1. VM has FirmwareUpToDate=False.
2. If KVVMI exists and is not Running, firmware updater does not create firmware-update-* VMOP.
3. If KVVMI is missing (NotFound) and other conditions are met, migration flow continues.
4. If KVVMI Get returns non-NotFound error, handler returns error.

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: "Run firmware workload-update migration only for running virtual machine instances, continue on missing KVVMI, and avoid unnecessary failed VMOPs for non-running VMs."
impact_level: low