feat(usage): 管理员用量统计支持筛选刷新与失败保留旧数据

- 用量统计/聚合接口新增 skipCache 选项与 120s 超时，便于强制绕过缓存
- loadStats 增加 force/preserveOnFailure 选项，背景刷新失败时保留旧数据
- 管理员页面在用户/模型/Provider 筛选变化时强制刷新统计，并将筛选条件传入统计接口
- 手动刷新与自动刷新分离，自动刷新不再重载长期热力图相关聚合

frontend/src/features/usage/composables/__tests__/useUsageData.spec.ts

@@ -219,4 +219,51 @@ describe('useUsageData', () => {
     expect(providerStats.value.map(item => item.provider)).toEqual(['OpenAI'])
     expect(availableProviders.value).toEqual(['OpenAI'])
   })
+
+  it('keeps previous admin provider stats when a background refresh fails', async () => {
+    const isAdminPage = ref(true)
+    const { loadStats, providerStats, availableProviders } = useUsageData({ isAdminPage })
+    const dateRange = { preset: 'last7days', tz_offset_minutes: 0 }
+
+    getUsageStatsMock.mockResolvedValueOnce({
+      total_requests: 3,
+      total_tokens: 300,
+      total_cost: 1,
+      avg_response_time: 0,
+    })
+    getUsageByProviderMock.mockResolvedValueOnce([
+      {
+        provider: 'OpenAI',
+        request_count: 3,
+        total_tokens: 300,
+        total_cost: 1.23,
+        actual_cost: 1.5,
+        avg_response_time_ms: 1250,
+        success_rate: 100,
+      },
+    ])
+
+    await loadStats(dateRange)
+
+    getUsageStatsMock.mockResolvedValueOnce({
+      total_requests: 4,
+      total_tokens: 400,
+      total_cost: 2,
+      avg_response_time: 0,
+    })
+    getUsageByProviderMock.mockRejectedValueOnce({
+      response: { status: 500 },
+      message: 'provider aggregation failed',
+    })
+
+    const hadFailure = await loadStats(dateRange, { preserveOnFailure: true })
+
+    expect(hadFailure).toBe(true)
+    expect(providerStats.value).toHaveLength(1)
+    expect(providerStats.value[0]).toMatchObject({
+      provider: 'OpenAI',
+      requests: 3,
+    })
+    expect(availableProviders.value).toEqual(['OpenAI'])
+  })
 })