The new CPU load metrics:
- kubevirt_vmi_guest_load_15m
- kubevirt_vmi_guest_load_1m
- kubevirt_vmi_guest_load_5m


The metrics were added to the `v1.VirtualMachineInstanceGuestAgentInfo` struct.

That struct is also used by `GetGuestInfo` for the cmd API (handler <-> launcher). This API is using `v1.VirtualMachineInstanceGuestAgentInfo` as response.

You can see it by running virtctl guestosinf.

So whatever you expose to  `v1.VirtualMachineInstanceGuestAgentInfo` is exposed by metrics and Subresource API.

As a result, when adding the CPU load metrics to that struct they were added to the API by mistake.

We should move the `Load` out of `v1.VirtualMachineInstanceGuestAgentInfo` to a different struct.

1.
2.
3.

CPU guest load metrics are reported in the API. Can be seen by running `virtctl guestosinf`.

The metrics should not be reported as an API.