M4b：ai_provider 删除手术①-⑥——神识全面接管模型出口

- ① 后端删除 app/ai_provider/（38 文件，含 anthropic/openai 格式、billing、audio、embedding）、services/ai_provider_admin.py、core/config/model_config.py、tests/ai_provider/（21 文件）
- ② 连带退役 application/agent/（旧 Agent 循环/steer/follow-up/确认注册表）、application/skill_executor.py、app/skills/（python skills 孤儿）、endpoints/audio.py、application/imports.py（ProviderManager 消费方）；settings 端点收敛为生成偏好
- ③ runs.py 修剪：AgentRunRegistry/steer/follow-up/confirmations 端点退役，保留 list/get/events/cancel/stream；safety 确认改为直接落库裁决（pending→approved/denied）
- ④ 前端删除：models 页、ModelGatewayView、gateway.ts、GatewayPanel（玄关）、ModelGatewayPanel/ModelStrategyForm、audio.ts/useVoiceInput、网关类型与 mock；设置页收敛为外观/语言/通知
- ⑤ 配置收敛：config/models|providers|pricing.example.yaml 归档 shenshi/config/；.env.example 删 provider keys/MODEL_ENDPOINT_POOL/MODEL_CONFIG_PATH，收敛 SHENSHI_*
- ⑥ 验收：rg 审计灵台 Provider SDK import 与 ai_provider 引用归零；后端 528 passed（2 个预存失败经 HEAD 复跑验证与本手术无关）、前端 type-check+201 passed、神识 go vet+test、BFF 4 passed

frontend/src/lib/mock/index.ts

@@ -64,12 +64,6 @@ function paginate<T>(items: T[]) {
   };
 }
 
-/** mock 内存态：模型 endpoint 覆盖（模拟 DB 覆盖，refresh 重置） */
-const _mockModelOverrides: Record<
-  string,
-  Record<string, Record<string, unknown>>
-> = {};
-
 /**
  * 路由匹配：命中返回 MockResult，未命中返回 undefined。
  * @param method HTTP 方法
@@ -88,54 +82,6 @@ export function resolveMock(
   // —— 健康检查 ——
   if (m === "GET" && path === "/health") return ok(data.health);
 
-  // —— 设置：模型网关状态（GET /settings/models，provider 密钥脱敏 + endpoint 覆盖标记）——
-  if (m === "GET" && path === "/settings/models") {
-    const gw = JSON.parse(JSON.stringify(data.modelGateway));
-    for (const model of gw.models) {
-      for (const ep of model.endpoints ?? []) {
-        const ov = _mockModelOverrides[model.name]?.[ep.name];
-        ep.priority = ov?.priority ?? ep.priority ?? 10;
-        ep.base_url_env = `${(ep.provider ?? "").toUpperCase()}_BASE_URL`;
-        ep.api_key_env = `${(ep.provider ?? "").toUpperCase()}_API_KEY`;
-        ep.is_overridden = Boolean(ov);
-        ep.has_url_override = Boolean(ov && ov.base_url);
-        ep.has_key_override = Boolean(ov && ov.api_key);
-        if (ov) {
-          if (ov.status) ep.status = ov.status;
-          if (ov.weight !== undefined) ep.weight = ov.weight;
-        }
-      }
-    }
-    return ok(gw);
-  }
-  {
-    const mt = path.match(/^\/settings\/models\/([^/]+)\/endpoints\/([^/]+)$/);
-    if (mt && m === "PATCH") {
-      const model = mt[1];
-      const endpoint = mt[2];
-      const body = (json ?? {}) as Record<string, unknown>;
-      _mockModelOverrides[model] = _mockModelOverrides[model] || {};
-      _mockModelOverrides[model][endpoint] = {
-        ...(_mockModelOverrides[model][endpoint] ?? {}),
-        ...body,
-      };
-      const ov = _mockModelOverrides[model][endpoint];
-      // 返回脱敏 endpoint（不回 base_url/api_key 原值）
-      return ok({
-        name: endpoint,
-        provider: "",
-        status: (ov.status as string | undefined) ?? "active",
-        weight: ov.weight as number | undefined,
-        priority: ov.priority as number | undefined,
-        is_overridden: true,
-        has_url_override: Boolean(ov.base_url),
-        has_key_override: Boolean(ov.api_key),
-        base_url_env: "",
-        api_key_env: "",
-      });
-    }
-  }
-
   // —— 设置：生成参数偏好（GET / PATCH /settings/preferences）——
   if (m === "GET" && path === "/settings/preferences") {
     return ok(data.generationPreferences);