feat(skills,knowledge): 中文名补全与知识库沉淀支持可选大模型并明确大模型报错

技能中文名:
- 单个「AI 生成」与「一键补充中文名」均可在请求体 model_name 选择大模型
- 一键补充改为仅管理员开放
- _generate_skill_name_zh 在模型自身失败(构建/超时/API 报错)时抛
  SkillNameZhModelError,与模型返回无效名称区分;单个接口返回
  502「大模型报错,无法补全技能中文名」,批量在 model_error 中带回代表性报错

知识库沉淀:
- 对话保存到知识库 / 批量导入均新增可选 model_name(透传至 capture_thread
  → extract_thread_llm_multi)
- 批量返回 llm_fallback 计数;模型异常时仍按既有逻辑回退规则抽取并落草稿,
  前端明确提示「已改用规则抽取」(保存按钮保持后台异步)

前端新增 components/workspace/model-name-select.tsx 复用下拉(默认=系统配置模型)

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

frontend-web/src/core/knowledge/types.ts

@@ -101,6 +101,8 @@ export interface IngestThreadPayload {
   include_sources?: boolean;
   /** When true the backend sediments asynchronously and returns immediately. */
   background?: boolean;
+  /** Optional model override for knowledge extraction (null = configured default). */
+  model_name?: string | null;
 }
 
 export interface BatchIngestPayload {
@@ -110,6 +112,8 @@ export interface BatchIngestPayload {
   include_sources?: boolean;
   skip_existing?: boolean;
   mode?: "summary" | "full";
+  /** Optional model override for knowledge extraction (null = configured default). */
+  model_name?: string | null;
 }
 
 export interface BatchIngestResult {
@@ -118,7 +122,14 @@ export interface BatchIngestResult {
   created: number;
   skipped: number;
   failed: number;
-  items: Array<{ thread_id: string; note_id: string | null; status: string }>;
+  /** Notes that fell back to rule-based extraction because the model failed. */
+  llm_fallback?: number;
+  items: Array<{
+    thread_id: string;
+    note_id: string | null;
+    status: string;
+    llm_fallback?: boolean;
+  }>;
 }
 
 export interface GraphExportResult {