Sync cezen-portal/ from live working copy (V1.0 GA polish + search-icon ordering fix)

Brings the ISO-packaged portal copy current with the live cezen-portal/
working directory, which had drifted across the entire V1.0 GA polish
initiative (Phases 1-10) plus this session's topnav icon-order bugfix.

- 52 existing files updated, 2 new pages added (about.html, health.html)
- cgit/cezen-portal/ is now byte-identical to cezen-portal/
- Backend files (rag_ingest.py, cezen_license.py) already in sync;
  main.py intentionally left as-is (cgit is ahead there with
  workstation-category commercial-fit logic not yet backported to the
  live working copy)

Next: rebuild the ISO (bash autoinstall/build-iso.sh) to bake this in.

cezen-portal/model-compare.html

@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <title>Model Comparison — Nexus One AI</title>
-  <link rel="stylesheet" href="style.css?v=4">
+  <link rel="stylesheet" href="style.css?v=12">
   <style>
     :root {
       --navy2: var(--surface);
@@ -258,6 +258,7 @@
         <a href="dashboard.html">Dashboard</a>
         <a href="analytics.html">Usage Analytics</a>
         <a href="audit.html">Governance & Audit</a>
+        <a href="health.html">Health Center</a>
         <a href="feedback.html">Feedback &amp; Ratings</a>
         <span class="nav-drop-cat">MANAGE /</span>
         <a href="users.html">Users</a>
@@ -297,7 +298,7 @@
       </div>
     </div>
   </nav>
-    <a href="notifications.html" style="position:relative">🔔</a>
+    <a href="notifications.html" class="nav-icon-btn" style="position:relative" aria-label="Notifications"><span class="icon-svg icon-sm" aria-hidden="true"><svg viewBox="0 0 24 24"><path d="M15 17h5l-1.4-1.4A2 2 0 0 1 18 14.2V11a6 6 0 1 0-12 0v3.2a2 2 0 0 1-.6 1.4L4 17h5"/><path d="M10 17a2 2 0 0 0 4 0"/></svg></span></a>
     <span class="badge" data-brand="tier">Basic Tier</span>
   <div id="nav-org-logo" class="nav-org-logo"></div>
 </header>
@@ -573,10 +574,10 @@ Ultra-fast response from the smallest model. Best for simple tasks where speed i
 Quick answer: The 3B model handles straightforward Q&A, classification, and short-form generation efficiently.
 
 **Trade-offs vs larger models:**
-- ✅ Fastest inference time
-- ✅ Lowest memory footprint (~2GB VRAM)
-- ⚠️ Less nuanced reasoning
-- ⚠️ Shorter effective context window
+- Fastest inference time
+- Lowest memory footprint (~2GB VRAM)
+- Less nuanced reasoning
+- Shorter effective context window
 
 Ideal for: chatbots, quick lookups, real-time classification.`
 };
@@ -766,12 +767,6 @@ function renderMd(raw) {
   return `<p>${html}</p>`;
 }
 
-function showToast(msg) {
-  const t = document.getElementById('mc-toast');
-  t.textContent = msg; t.classList.add('show');
-  setTimeout(() => t.classList.remove('show'), 2200);
-}
-
 // ── Init ──────────────────────────────────────────────────────────────────────
 // Load import from compare button in other pages
 const compareImport = localStorage.getItem('cezen_compare_import');
@@ -784,6 +779,6 @@ if (compareImport) {
 </script>
 
 <script src="auth.js"></script>
-<script src="branding.js"></script>
+<script src="branding.js?v=3"></script>
 </body>
 </html>