docs(architecture): 完成技术验证 S1–S4，确定 FR-M2-05/06 目标值

- spikes/：Word 转 PDF（Gotenberg）、格式识别、原句坐标、DeepSeek 抽取四项验证代码与两版提示词；样本派生产物和 .env 不进 git
- tech-validation.md：完整结论。四项都可行：WPS 字体字符集要修正并改用行高调整过的字体；格式识别不需要 Docling；表格内切句待改进；抽取需第三道依据校验与规则兜底
- architecture.md：§3.5 加预处理与字体包，§3.6 证据校验改为三道，§10 回填结论
- 03 功能需求与 decisions.md（TV1）：召回 ≥ 99%、采纳准确率 ≥ 95%、页码 100%、高亮正文 ≥ 98% / 表格 ≥ 95%

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

spikes/s1-convert/truth.py

@@ -0,0 +1,32 @@
+"""S1 真值：Word 最后保存时记录的页数。DOCX 读 docProps/app.xml 的 <Pages>，旧版 DOC 读 OLE SummaryInformation 的 PIDSI_PAGECOUNT。"""
+import json, re, sys, zipfile, pathlib
+sys.path.insert(0, str(pathlib.Path(__file__).resolve().parent.parent))
+import olefile
+from common import IDS, sample_file, sniff
+
+OUT = pathlib.Path(__file__).parent / "out"
+rows = {}
+for sid in IDS:
+    f = sample_file(sid)
+    kind = sniff(f)
+    info = {"kind": kind, "pages": None, "app": None, "ole_objects": 0}
+    if kind == "docx":
+        with zipfile.ZipFile(f) as z:
+            names = z.namelist()
+            if "docProps/app.xml" in names:
+                app = z.read("docProps/app.xml").decode("utf-8", "replace")
+                m = re.search(r"<Pages>(\d+)</Pages>", app)
+                info["pages"] = int(m.group(1)) if m else None
+                a = re.search(r"<Application>([^<]*)</Application>", app)
+                info["app"] = a.group(1) if a else None
+            info["ole_objects"] = sum(1 for n in names if n.startswith("word/embeddings/"))
+            info["has_ole_xml"] = b"<o:OLEObject" in z.read("word/document.xml")
+    elif kind == "doc":
+        ole = olefile.OleFileIO(str(f))
+        meta = ole.get_metadata()
+        info["pages"] = meta.num_pages
+        info["app"] = (meta.creating_application or b"").decode("gbk", "replace") if isinstance(meta.creating_application, bytes) else meta.creating_application
+        info["ole_objects"] = sum(1 for e in ole.listdir() if e and e[0] == "ObjectPool")
+    rows[sid] = info
+    print(sid, info)
+(OUT / "truth.json").write_text(json.dumps(rows, ensure_ascii=False, indent=1))