R-772: a health probe that could not run records not_checked + NOT healthy, is looked at again on the 10 s cycle, and the app page says so

The state stays the containers' (probeSaysUnhealthy skips a not-checked record), so R-630 holds.
Red-proof RP-D1 — felhom.eu audits/visitors-2026-10-01/D.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS

controller/internal/stacks/healthprobe.go

@@ -82,13 +82,17 @@ func (m *Manager) RunHealthProbes() error {
 	for _, np := range noProbe {
 		m.mu.Lock()
 		if s, ok := m.stacks[np.name]; ok {
+			// R-772: NOT healthy — no check ran. R-630's point still holds through NotChecked: the state is read from
+			// the containers, never forced to unhealthy by this record. Not healthy also means the 10-second fast
+			// cycle, so a container that comes back is checked within seconds, not 5 minutes later.
 			s.HealthProbe = &HealthProbeResult{
-				Healthy:   true, // never RED on our own inability to look — R-630's whole point
-				LastCheck: m.now(),
+				Healthy:    false,
+				NotChecked: true,
+				LastCheck:  m.now(),
 				Details: []HealthCheckDetail{{
 					Type:       "none",
 					Target:     np.name,
-					Healthy:    true,
+					Healthy:    false,
 					Error:      MsgHealthNoProbeContainer,
 					MessageKey: KeyHealthNoProbeContainer,
 				}},