[glm-grade=A] feat: add disk-safety warning to setup wizard + health retention settings

C-grade round-1 blockers fixed:
- [HIGH] retention default 14d → 30d to match engine (health-checker.js:34)
- [MEDIUM] phantom 'Settings → Disk Safety' path removed
- [MEDIUM] dangling 'stats polling interval' bullet (no such control in modal)
- [LOW] exaggerated 'hundreds of MB' → 'tens of MB'
- [LOW] button label mismatch (real button is '💾 Disk')

GLM round 2 verified all 5 fixes landed; no new regressions; HTML balanced.

Pre-existing follow-up parked: disk-settings.json saved values not reloaded by engine on container restart (out of scope for this commit).

status/js/disk-settings.js

@@ -47,6 +47,15 @@
           </div>
         </div>
 
+        <div style="background:rgba(243,156,18,0.08);border:1px solid rgba(243,156,18,0.25);border-radius:8px;padding:12px;margin-bottom:16px;">
+          <div style="font-size:0.82rem;color:#f0a040;line-height:1.45;">
+            ⚠ <strong>Disk impact:</strong> Lowering the health check interval increases how often data is written to disk.
+            DashCaddy caps history at <strong>max entries per service</strong> and prunes entries older than the retention period,
+            so these two values together determine steady-state disk usage. For busy hosts, prefer a longer interval (60–120s)
+            and a lower entry cap.
+          </div>
+        </div>
+
         <div style="display:grid;gap:16px;">
           ${settingRow('Health Check Interval', 'healthInterval', c.healthCheckInterval, 'seconds', Math.round((c.healthCheckInterval||30000)/1000), 5, 300)}
           ${settingRow('Max Health Entries/Service', 'healthMaxEntries', c.healthMaxEntries, 'entries', c.healthMaxEntries||500, 50, 5000)}