fix(boundary): 修复多库边界问题、版本管理及删除清理

多库检索与存储修复:
- RRF 融合去重改用 (collection, chunk_id) 复合键，修复同名文件结果被吞
- DocStore 存储路径加 collection 前缀，修复跨库同名切片数据覆盖
- search_multiple 去重改用复合键
- chunk_id 解析改用 rsplit 兼容下划线文件名

上传与版本管理修复:
- 同名文件上传改为覆盖模式，自动清理旧切片
- 修复首次上传不创建版本记录
- 修复覆盖上传版本号回退到 v1
- sync ADDED 分支改用动态版本号生成
- _generate_version_id 改为基于全部版本递增
- 废止/恢复操作同步 SQLite 版本记录
- mark_document_as_superseded 改为仅更新 SQLite

删除清理修复:
- 删除文档时同步清理 SQLite 版本记录和变更日志
- 删除向量库时同步清理该库所有版本记录
- cleanup 改为清理 SQLite 记录而非 ChromaDB

测试:
- test_version_management.py: 27 条版本管理单元测试
- test_edge_cases.py: 28 条边界用例测试
- test_upload_dedup.py: 5 条上传去重测试
- e2e_risk_test.py: 27 条端到端风险测试

文档:
- 新增风险边界问题修复注意事项.md（面向后端的对接文档）
- 新增向量库边界风险分析.md
- 更新多篇现有文档

api/document_routes.py

@@ -243,14 +243,75 @@ def upload_document() -> Tuple[Any, int]:
         filename = f"upload_{datetime.now().strftime('%Y%m%d%H%M%S')}{ext}"
 
     filepath = os.path.join(target_dir, filename)
-    if os.path.exists(filepath):
-        timestamp = datetime.now().strftime('_%Y%m%d_%H%M%S')
-        name, ext_part = os.path.splitext(filename)
-        filename = f"{name}{timestamp}{ext_part}"
-        filepath = os.path.join(target_dir, filename)
 
+    # 同名文件处理：清理旧切片 + 覆盖旧文件（替代原来的时间戳重命名策略）
+    replaced = False
+    if os.path.exists(filepath):
+        replaced = True
+        # 1. 清理旧切片（ChromaDB + BM25 + DocStore）
+        kb_manager = _get_kb_manager()
+        if kb_manager:
+            old_chunks = kb_manager.get_document_chunks(collection, filename)
+            if old_chunks:
+                old_ids = [c['id'] for c in old_chunks]
+                coll_obj = kb_manager.get_collection(collection)
+                if coll_obj:
+                    coll_obj.delete(ids=old_ids)
+                    logger.info(f"替换上传: 清理旧切片 {filename} -> {collection}, 共 {len(old_ids)} 个")
+
+                    # 清理关联的 DocStore 文件
+                    try:
+                        from pathlib import Path as _Path
+                        docstore_dir = _Path('.data/docstore')
+                        if docstore_dir.exists():
+                            for ds_file in docstore_dir.glob(f'{collection}_{filename}_*.json'):
+                                ds_file.unlink()
+                    except Exception as e:
+                        logger.warning(f"清理 DocStore 失败: {e}")
+
+                    # 重建 BM25 索引
+                    kb_manager.rebuild_bm25_index(collection)
+
+        # 2. 清理同步哈希记录（以便 sync 能正确检测为新文件）
+        try:
+            from knowledge.sync import SyncDatabase
+            sync_db = SyncDatabase()
+            sync_db.delete_document_hash(f"{collection}/{filename}")
+        except Exception as e:
+            logger.warning(f"清理哈希记录失败: {e}")
+
+    # 保存文件（覆盖同名旧文件）
     file.save(filepath)
 
+    # 版本管理：覆盖上传时标记旧版本为 superseded（新版本记录由 sync 服务统一创建）
+    if replaced:
+        try:
+            from knowledge.document_versions import get_version_query
+            from data.db import get_connection
+            vq = get_version_query()
+            active = vq.get_active_version(collection, filename)
+
+            if active:
+                with get_connection("knowledge") as conn:
+                    conn.execute("""
+                        UPDATE document_versions
+                        SET status='superseded', deprecated_date=?,
+                            deprecated_reason='重新上传覆盖'
+                        WHERE collection=? AND document_id=? AND version=?
+                    """, (datetime.now().isoformat(), collection,
+                          filename, active.version))
+                    conn.commit()
+                vq.log_version_change(
+                    collection, filename,
+                    change_type="reupload",
+                    old_version=active.version,
+                    old_status="active", new_status="superseded",
+                    reason="重新上传覆盖",
+                    changed_by=request.current_user.get('user_id', '')
+                )
+        except Exception as e:
+            logger.warning(f"标记旧版本失败: {e}")
+
     # 6. 触发向量化
     sync_status = "已保存，等待手动同步"
     sync_service = _get_sync_service()
@@ -276,7 +337,8 @@ def upload_document() -> Tuple[Any, int]:
                 "filename": filename,
                 "collection": collection,
                 "path": f"{target_subdir}/{filename}",
-                "size": file_size
+                "size": file_size,
+                "replaced": replaced
             },
             "sync_status": sync_status
         },
@@ -350,19 +412,56 @@ def batch_upload_documents() -> Tuple[Any, int]:
                 filename = f"upload_{datetime.now().strftime('%Y%m%d%H%M%S')}{ext}"
             filepath = os.path.join(target_dir, filename)
 
-            # 处理重名
+            # 同名文件处理：清理旧切片 + 覆盖旧文件
+            replaced = False
             if os.path.exists(filepath):
-                timestamp = datetime.now().strftime('_%Y%m%d_%H%M%S')
-                name, ext_part = os.path.splitext(filename)
-                filename = f"{name}{timestamp}{ext_part}"
-                filepath = os.path.join(target_dir, filename)
+                replaced = True
+                kb_mgr = _get_kb_manager()
+                if kb_mgr:
+                    old_chunks = kb_mgr.get_document_chunks(collection, filename)
+                    if old_chunks:
+                        old_ids = [c['id'] for c in old_chunks]
+                        coll_obj = kb_mgr.get_collection(collection)
+                        if coll_obj:
+                            coll_obj.delete(ids=old_ids)
+                            logger.info(f"批量替换: 清理旧切片 {filename} -> {collection}, 共 {len(old_ids)} 个")
+                    kb_mgr.rebuild_bm25_index(collection)
+
+                try:
+                    from knowledge.sync import SyncDatabase
+                    sync_db = SyncDatabase()
+                    sync_db.delete_document_hash(f"{collection}/{filename}")
+                except Exception as e:
+                    logger.warning(f"清理哈希记录失败: {e}")
 
             file.save(filepath)
 
+            # 版本管理：覆盖上传时标记旧版本为 superseded（新版本记录由 sync 服务统一创建）
+            if replaced:
+                try:
+                    from knowledge.document_versions import get_version_query
+                    from data.db import get_connection
+                    vq = get_version_query()
+                    active = vq.get_active_version(collection, filename)
+
+                    if active:
+                        with get_connection("knowledge") as conn:
+                            conn.execute("""
+                                UPDATE document_versions
+                                SET status='superseded', deprecated_date=?,
+                                    deprecated_reason='批量上传覆盖'
+                                WHERE collection=? AND document_id=? AND version=?
+                            """, (datetime.now().isoformat(), collection,
+                                  filename, active.version))
+                            conn.commit()
+                except Exception as e:
+                    logger.warning(f"批量上传标记旧版本失败: {e}")
+
             results.append({
                 "filename": filename,
                 "status": "success",
-                "path": f"{target_subdir}/{filename}"
+                "path": f"{target_subdir}/{filename}",
+                "replaced": replaced
             })
         except Exception as e:
             results.append({