feat: 流式空闲超时、健康监控查询优化、限流桶内存上限与维护清理修复

Close #233

Co-authored-by: AAEE86 <ppk0227@hotmail.com>

- cli_monitor_mixin: 引入 STREAM_IDLE_TIMEOUT_SECONDS（可通过环境变量配置），
  流传输开始后若超出空闲窗口无新 chunk 则提前取消并返回 504，避免长时间挂起
- stream_context: 新增 managed_recorded_bodies 上下文管理器，确保 chunks 在
  telemetry 完成后及时释放；stream_telemetry 使用该接口统一管理 response body 构建
- health endpoint: 将状态聚合改为 GROUP BY 直接统计，事件列表按 api_format
  单独查询，避免单次 limit 拉取大量记录导致的遗漏与性能问题；同时过滤不活跃
  provider/endpoint，与公开健康接口保持一致
- endpoint health service: 修正时间线数据按 endpoint_id 而非 key_id 聚合
- token_bucket: 引入 max_buckets/bucket_expiry 上限与定时清理，防止内存无限增长；
  修复 refill_rate=0 时 get_reset_time 除零异常；新增 _is_unlimited_rate_limit 判断
- maintenance_scheduler: 调整清理顺序（先删整行再按窗口清理），新增 newer_than
  边界参数，避免同一行在同一轮中被重复改写
- sync_execute: 新增 create_pending_usage 开关，允许已预创建记录的调用方跳过重复创建
- quota_reader / provider_ops balance: 小幅修复与健壮性提升
- Dockerfile: 添加 MALLOC_ARENA_MAX=2 环境变量以降低 gunicorn worker RSS
- 补充相关测试覆盖

frontend/src/composables/useCountdownTimer.ts

@@ -84,15 +84,24 @@ export interface CodexResetStatus {
  * @param resetSecs 相对剩余秒数（用于 fallback）
  * @param updatedAt 元数据更新时间（Unix 秒）
  * @param _tick 响应式触发器（传入 tick.value 以触发响应式更新）
+ * @param remainingPercent 当前窗口剩余额度百分比（0-100，100 表示满额不启动倒计时）
  */
 export function getCodexResetCountdown(
   resetAt: number | null | undefined,
   resetSecs: number | null | undefined,
   updatedAt: number | null | undefined,
-  _tick: number
+  _tick: number,
+  remainingPercent?: number | null
 ): CodexResetStatus | null {
   void _tick
 
+  if (remainingPercent != null) {
+    const normalizedRemaining = Number(remainingPercent)
+    if (Number.isFinite(normalizedRemaining) && normalizedRemaining >= 100) {
+      return null
+    }
+  }
+
   const nowSec = Math.floor(Date.now() / 1000)
   let remaining: number
 

frontend/src/features/providers/components/ProviderDetailDrawer.vue

@@ -573,18 +573,20 @@
                             />
                           </div>
                           <div
-                            v-if="key.upstream_metadata.codex.primary_reset_at || key.upstream_metadata.codex.primary_reset_seconds"
+                            v-if="(key.upstream_metadata.codex.primary_reset_at || key.upstream_metadata.codex.primary_reset_seconds) && shouldStartCodexResetCountdown(key.upstream_metadata.codex.primary_used_percent)"
                             class="text-[9px] mt-0.5 tabular-nums"
                             :class="getResetCountdownClass(
                               key.upstream_metadata.codex.primary_reset_at,
                               key.upstream_metadata.codex.primary_reset_seconds,
-                              key.upstream_metadata.codex.updated_at
+                              key.upstream_metadata.codex.updated_at,
+                              key.upstream_metadata.codex.primary_used_percent
                             )"
                           >
                             {{ getResetCountdownText(
                               key.upstream_metadata.codex.primary_reset_at,
                               key.upstream_metadata.codex.primary_reset_seconds,
-                              key.upstream_metadata.codex.updated_at
+                              key.upstream_metadata.codex.updated_at,
+                              key.upstream_metadata.codex.primary_used_percent
                             ) }}
                           </div>
                         </div>
@@ -604,18 +606,21 @@
                             />
                           </div>
                           <div
+                            v-if="shouldStartCodexResetCountdown(key.upstream_metadata.codex.secondary_used_percent)"
                             class="text-[9px] mt-0.5 tabular-nums"
                             :class="getResetCountdownClass(
                               key.upstream_metadata.codex.secondary_reset_at,
                               key.upstream_metadata.codex.secondary_reset_seconds,
-                              key.upstream_metadata.codex.updated_at
+                              key.upstream_metadata.codex.updated_at,
+                              key.upstream_metadata.codex.secondary_used_percent
                             )"
                           >
                             <template v-if="key.upstream_metadata.codex.secondary_reset_at || key.upstream_metadata.codex.secondary_reset_seconds">
                               {{ getResetCountdownText(
                                 key.upstream_metadata.codex.secondary_reset_at,
                                 key.upstream_metadata.codex.secondary_reset_seconds,
-                                key.upstream_metadata.codex.updated_at
+                                key.upstream_metadata.codex.updated_at,
+                                key.upstream_metadata.codex.secondary_used_percent
                               ) }}
                             </template>
                             <template v-else>
@@ -2549,9 +2554,16 @@ function getAntigravityQuotaSummary(metadata: UpstreamMetadata | null | undefine
 function getResetCountdownText(
   resetAt: number | null | undefined,
   resetSecs: number | null | undefined,
-  updatedAt: number | null | undefined
+  updatedAt: number | null | undefined,
+  usedPercent: number | null | undefined
 ): string {
-  const status = getCodexResetCountdown(resetAt, resetSecs, updatedAt, countdownTick.value)
+  const status = getCodexResetCountdown(
+    resetAt,
+    resetSecs,
+    updatedAt,
+    countdownTick.value,
+    toCodexRemainingPercent(usedPercent)
+  )
   if (!status) return ''
   return status.isExpired ? status.text : `${status.text} 后重置`
 }
@@ -2559,15 +2571,35 @@ function getResetCountdownText(
 function getResetCountdownClass(
   resetAt: number | null | undefined,
   resetSecs: number | null | undefined,
-  updatedAt: number | null | undefined
+  updatedAt: number | null | undefined,
+  usedPercent: number | null | undefined
 ): string {
-  const status = getCodexResetCountdown(resetAt, resetSecs, updatedAt, countdownTick.value)
+  const status = getCodexResetCountdown(
+    resetAt,
+    resetSecs,
+    updatedAt,
+    countdownTick.value,
+    toCodexRemainingPercent(usedPercent)
+  )
   if (!status || status.isExpired) return 'text-muted-foreground/70'
   if (status.isCritical) return 'text-destructive font-medium animate-pulse'
   if (status.isUrgent) return 'text-amber-500 dark:text-amber-400'
   return 'text-muted-foreground/70'
 }
 
+function toCodexRemainingPercent(usedPercent: number | null | undefined): number | null {
+  const normalizedUsed = Number(usedPercent)
+  if (!Number.isFinite(normalizedUsed)) return null
+  const clampedUsed = Math.min(Math.max(normalizedUsed, 0), 100)
+  return Math.max(100 - clampedUsed, 0)
+}
+
+function shouldStartCodexResetCountdown(usedPercent: number | null | undefined): boolean {
+  const remainingPercent = toCodexRemainingPercent(usedPercent)
+  if (remainingPercent == null) return true
+  return remainingPercent < 100
+}
+
 // 格式化重置时间
 function formatResetTime(seconds: number): string {
   const days = Math.floor(seconds / 86400)

frontend/src/views/admin/PoolManagement.vue

@@ -2428,7 +2428,7 @@ function getQuotaProgressLabel(label: string): string {
 
 function getQuotaProgressCountdown(item: QuotaProgressItem) {
   if ((item.label !== '5H' && item.label !== '周') || item.resetAtSeconds == null) return null
-  return getCodexResetCountdown(item.resetAtSeconds, null, null, countdownTick.value)
+  return getCodexResetCountdown(item.resetAtSeconds, null, null, countdownTick.value, item.remainingPercent)
 }
 
 function getQuotaProgressCountdownText(item: QuotaProgressItem): string {
@@ -2438,6 +2438,9 @@ function getQuotaProgressCountdownText(item: QuotaProgressItem): string {
 }
 
 function getQuotaProgressTooltip(item: QuotaProgressItem): string {
+  if ((item.label === '5H' || item.label === '周') && item.remainingPercent >= 100) {
+    return ''
+  }
   const detail = item.detail?.trim() || ''
   const countdownText = getQuotaProgressCountdownText(item)
   if (countdownText) {