Fix M-4: punctuation-only page name yields empty slug ('.html')

slugify('!!!') stripped down to '' since punctuation-only names have no
a-z0-9 characters left; buildSavePayload then wrote filename = '' +
'.html'. slugify now falls back to 'page' when the computed slug is
empty, so uniqueSlug's existing dedupe (page, page-2, ...) applies same
as any other collision. Landing page slug stays forced to 'index'.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

craft/src/state/PageContext.slug.test.tsx

@@ -121,6 +121,38 @@ describe('PageProvider slug dedupe', () => {
     unmount();
   });
 
+  test('M-4: a punctuation-only page name ("!!!") falls back to slug "page", never empty string', () => {
+    let ctx: ReturnType<typeof usePages> | null = null;
+    const Consumer: React.FC = () => {
+      ctx = usePages();
+      return null;
+    };
+
+    render(
+      <PageProvider>
+        <Consumer />
+      </PageProvider>,
+    );
+
+    act(() => {
+      ctx!.addPage('!!!', '');
+    });
+
+    const added = ctx!.pages.find((p) => p.name === '!!!')!;
+    expect(added.slug).toBe('page');
+    expect(added.slug).not.toBe('');
+
+    // A second punctuation-only-named page dedupes to 'page-2', not ''.
+    act(() => {
+      ctx!.addPage('???', '');
+    });
+    const second = ctx!.pages.find((p) => p.name === '???')!;
+    expect(second.slug).toBe('page-2');
+    expect(second.slug).not.toBe('');
+
+    unmount();
+  });
+
   test('landing page slug always stays "index" even if renamed to collide', () => {
     let ctx: ReturnType<typeof usePages> | null = null;
     const Consumer: React.FC = () => {