Fixed
- Bogus "turned UNHEALTHY (health was: unhealthy)" alerts (#2, @famewolf). The monitor could fire a nonsensical unhealthy alert in two cases: for a container that was already unhealthy when Docksentry built its baseline (a restart rebuilds the baseline), and for stopped containers — a stopped container keeps its last
Health.Statusfrozen indocker inspect, and the monitor was treating that stale value as a live signal (a months-dead container triggered it). Now an unhealthy alert only fires for a healthy→unhealthy transition Docksentry actually observed on a running container, so the impossible "was: unhealthy" message can't happen and a pre-existing or long-dead container stays quiet.