feat(search): index PDF text, OCR scanned PDFs on import

Search now covers handwriting, the PDF text layer, AND scanned
(rasterized) PDFs.

- PdfTextIndexer runs at import: sums the embedded text layer across
  pages; if present it stores that as the document body, otherwise the
  PDF is rasterized and its rendered pages are OCR'd in the background.
  The result lands in the sidecar `pageText` field (distinct from
  `ocrText`, the handwriting OCR). Idempotent (skips a sidecar that
  already has pageText); degrades gracefully with no OCR engine.
- pdfrx_page_text_source abstracts text/render so it's testable.
- VaultSearchIndex now harvests title + typed text + handwriting OCR +
  PDF pageText, so search finds notes, typed PDFs and scanned PDFs.

analyze clean, 409 tests green.

test/vault_search_index_test.dart

@@ -8,8 +8,9 @@
 //   * the title / source filename,
 //   * a typed text box (EditorStroke.textContent),
 //   * the persisted handwriting OCR text (sidecar `ocrText`),
+//   * the PDF document-body text captured at import (sidecar `pageText`): both
+//     the embedded text layer AND the background-OCR result for a scanned PDF,
 //   * a CJK substring (this user writes Chinese).
-// It also documents the known GAP: a PDF's embedded text layer is NOT indexed.
 
 import 'dart:io';
 
@@ -56,6 +57,7 @@ void main() {
     required String pdfName,
     List<EditorStroke> page0 = const [],
     String? ocrText,
+    String? pageText,
   }) async {
     final dir = Directory(p.join(vaultDir.path, folder));
     await dir.create(recursive: true);
@@ -66,6 +68,7 @@ void main() {
       docType: 'pdf',
       strokes: page0.isEmpty ? null : {0: page0},
       ocrText: ocrText,
+      pageText: pageText,
       createdAt: DateTime.now().toUtc(),
     );
     await SidecarStore.writeAtomic(
@@ -173,11 +176,39 @@ void main() {
     expect(await index.search('x'), isEmpty);
   });
 
-  test('KNOWN GAP: a PDF embedded text layer is NOT indexed', () async {
-    // The sidecar carries no annotations; only the PDF body would contain the
-    // word "bodytext". Search does NOT read the PDF text layer (documented
-    // limitation), so this returns nothing.
-    await seedDocNotebook(folder: 'Plain', pdfName: 'plain.pdf');
+  test('finds a PDF by its embedded text layer (sidecar pageText)', () async {
+    // The PDF body text captured at import lives in the sidecar's `pageText`.
+    // No annotations at all — only the document body contains "bodytext".
+    await seedDocNotebook(
+      folder: 'Plain',
+      pdfName: 'plain.pdf',
+      pageText: 'introduction to bodytext and more printed content',
+    );
+    final index = VaultSearchIndex(vault);
+    final hits = await index.search('bodytext');
+    expect(hits, hasLength(1));
+    expect(hits.single.entry.openPath, endsWith('plain.pdf'));
+  });
+
+  test('finds a SCANNED PDF by its background-OCR pageText (CJK)', () async {
+    // A rasterized/scanned PDF has no text layer; the import-time OCR pass
+    // writes the recognized text into the SAME `pageText` field, so search
+    // covers scanned documents — including Chinese substrings.
+    await seedDocNotebook(
+      folder: '扫描讲义',
+      pdfName: 'scanned.pdf',
+      pageText: '微积分第三讲 导数的定义与几何意义',
+    );
+    final index = VaultSearchIndex(vault);
+    expect(await index.search('导数'), hasLength(1));
+    expect((await index.search('几何')).single.entry.openPath,
+        endsWith('scanned.pdf'));
+  });
+
+  test('a PDF with no pageText (un-indexed) is not found by body text', () async {
+    // Back-compat: a PDF imported before the feature (or whose OCR backend was
+    // unavailable) has no `pageText`; only its annotations are searchable.
+    await seedDocNotebook(folder: 'Old', pdfName: 'old.pdf');
     final index = VaultSearchIndex(vault);
     expect(await index.search('bodytext'), isEmpty);
   });