Currently thrown exceptions (mostly CLI commands not returning expected text and applying aggregation) are only logged to the console and then forgotten.
Instead, we can raise another event (careful; to prevent infinite re-invocation of the event handler) which can be populated either by SignalR to the actual consuming client (ioBroker, ..) or via metrics and imported in grafana.
The latter approach is preferred as it integrates nicely with the dashboard. -> TBD: What detail to expose within the scraping endpoint