fix(exam-prep): normalize LaTeX in dashboard preview text

Recent-attempts widget on /exam-prep/:examKey was showing raw LaTeX
like '\dfrac{7}{9}' because stripPreview only removed HTML tags.
Now it also converts common LaTeX to readable unicode (fractions →
a/b, \sqrt → √, \cdot → ·, comparisons → ≤≥≠, Greek letters, etc.)
before truncating.

KaTeX rendering would be overkill for a 100-char preview row; this
just makes the existing text legible.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

backend/src/routes/exam-prep.js

@@ -796,13 +796,39 @@ router.get('/:examKey/dashboard', (req, res) => {
   });
 });
 
-/* Lightweight preview: strip tags, normalize whitespace, truncate. */
+/* Lightweight preview: strip HTML, normalize LaTeX to readable unicode,
+   collapse whitespace, truncate. Used in dashboard "recent attempts" rows
+   where rendering KaTeX would be overkill but raw `$\dfrac{7}{9}$` is
+   ugly. Matches the spirit of tag-exam-tasks.js#stripText. */
 function stripPreview(html) {
-  const text = String(html || '')
+  let text = String(html || '')
     .replace(/<svg[\s\S]*?<\/svg>/gi, '')
-    .replace(/<[^>]+>/g, ' ')
+    .replace(/<[^>]+>/g, ' ');
+
+  // Common LaTeX → unicode/plain. Order matters: handle fractions/roots
+  // before stripping bare backslash commands.
+  text = text
+    .replace(/\\d?frac\s*\{([^{}]+)\}\s*\{([^{}]+)\}/g, '$1/$2')
+    .replace(/\\sqrt\s*\{([^{}]+)\}/g, '√($1)')
+    .replace(/\\cdot/g, '·')
+    .replace(/\\times/g, '×')
+    .replace(/\\pm/g, '±')
+    .replace(/\\mp/g, '∓')
+    .replace(/\\leq?/g, '≤')
+    .replace(/\\geq?/g, '≥')
+    .replace(/\\neq/g, '≠')
+    .replace(/\\approx/g, '≈')
+    .replace(/\\infty/g, '∞')
+    .replace(/\\pi/g, 'π')
+    .replace(/\\alpha/g, 'α').replace(/\\beta/g, 'β').replace(/\\gamma/g, 'γ')
+    .replace(/\\(?:left|right)/g, '')
+    .replace(/\\[a-zA-Z]+\s*/g, ' ')   // any remaining LaTeX command
+    .replace(/[{}]/g, '')               // leftover braces
+    .replace(/\$+/g, '')                // math delimiters
+    .replace(/&[a-z]+;/gi, ' ')         // html entities
     .replace(/\s+/g, ' ')
     .trim();
+
   return text.length > 100 ? text.slice(0, 100) + '…' : text;
 }