When you create a lot of Droplets, LoadBalancers, etc. you possibly would like to have them grouped together - so that's where tags came.
Also Load Balancers spawned from Kubernetes Clusters (because of the usage of K8S service with type LoadBalancer) doesn't have any visible relationship to the "parent" cluster - tags similar to tags put on Nodes/Droplets would be useful - for example: k8s, k8s:cluster-id, k8s:service:k8s-service-name, etc.
Another usage example is that when you maintain your infrastructure by script you would like to have a place for keeping "metadata" about the droplet/loadbalancer/cluster/etc. And Load Balancers lack this feature because of no tags.