Currently in cnv-tests, when a VMI starts and encounters ErrorUnschedulable status, a failure is immediately declared.
This contradicts kubernetes flow, where ErrorUnschedulable is considered a temporary and recoverable state. We nede to implement this approach in network tests.
One scenario where this is useful is in L2 setups: When a bridhe interface is created on a cluster node, it might take some time for the bridge-marker to add a mark for this bridge in the node resource.
If a VM/pod that references this bridge (via NAD) is spinned-up while the amrker is not updated yet, the workload will reach ErrorUnschedulable and independently proceed once the marker is available.