Fix unnecessary keepalived switches due to unhealthy condition

[dergeberl]

In some cases, a change (like nodes or ports) triggered an unnecessary keepalived failover.

It happened that the envoy did not have the new snapshot version here:

yawol/controllers/yawollet/loadbalancer_controller.go

Lines 162 to 167 in 609b3e0

	// check envoy snapshot and update condition
	if changed {
	helper.CheckEnvoyVersion(conditions, &snapshot)
	} else {
	helper.CheckEnvoyVersion(conditions, oldSnapshot)
	}

This results in a false EnvoyUpToDate condition. That will not be fixed until the requeue time (30s) but the keepalived healthcheck of yawollet fails which results in an unnecessary keepalived failover.

I think the best way to solve this is to use the same check for unready conditions and return an error to requeue with backoff.