feat(website): add PDF to Markdown answers

.github/workflows/website-production-check.yml

@@ -22,11 +22,12 @@ jobs:
             /large-markdown-files/
             /markdown-to-slides/
             /convert-to-markdown/
+            /pdf-to-markdown/
             /scan-to-markdown/
             /bibtex-citations/
           )
           for language in zh-cn zh-tw ja; do
-            for suffix in / /workflows/ /download/ /large-markdown-files/ /markdown-to-slides/ /convert-to-markdown/ /scan-to-markdown/ /bibtex-citations/; do
+            for suffix in / /workflows/ /download/ /large-markdown-files/ /markdown-to-slides/ /convert-to-markdown/ /pdf-to-markdown/ /scan-to-markdown/ /bibtex-citations/; do
               routes+=("/${language}${suffix}")
             done
           done

showcase/reports/pdf_answer_pages_approval.json

@@ -0,0 +1,52 @@
+{
+  "schema_version": 1,
+  "artifact": "Localized PDF to Markdown answer pages",
+  "reviewed_at": "2026-08-26T01:20:00Z",
+  "required_rounds": 3,
+  "minimum_score": 9.7,
+  "rounds": [
+    {
+      "round": 1,
+      "reviewer": "seo-intent-review",
+      "status": "approved",
+      "score": 9.89,
+      "evidence": [
+        "The PDF to Markdown family targets a distinct high-intent task rather than duplicating the general conversion page.",
+        "All four localized pages expose unique titles, descriptions, canonical URLs, hreflang sets, FAQPage, HowTo, and BreadcrumbList entities.",
+        "Home, workflow, download, sitemap, llms.txt, and llms-full.txt surfaces link the new canonical paths."
+      ]
+    },
+    {
+      "round": 2,
+      "reviewer": "content-accuracy-review",
+      "status": "approved",
+      "score": 9.83,
+      "evidence": [
+        "The pages qualify text-based conversion separately from scanned OCR workflows instead of promising perfect extraction.",
+        "Privacy wording remains limited to the implemented local-conversion boundary and explicitly excludes cloud upload requirements.",
+        "Each language instructs the reader to review reading order, tables, captions, names, numbers, formulas, and complex layouts before saving."
+      ]
+    },
+    {
+      "round": 3,
+      "reviewer": "release-engineering-review",
+      "status": "approved",
+      "score": 9.79,
+      "evidence": [
+        "Website validator passes 36 canonical URLs, including four new localized PDF answers, with three recorded approval rounds.",
+        "Release build validation passes after the expanded corpus and sitemap are copied into dist.",
+        "Browser smoke passes all 36 routes with zero failures, zero broken images, and matching PNG social metadata."
+      ]
+    }
+  ],
+  "final_decision_meeting": {
+    "status": "approved_for_staged_publication",
+    "score": 9.84,
+    "participants": ["SEO intent", "content accuracy", "release engineering"],
+    "conditions": [
+      "CI must pass before merge.",
+      "Production check must be updated to include all twelve localized PDF route variants before deployment.",
+      "IndexNow resubmission must follow a successful production check."
+    ]
+  }
+}