perf(queue): 限制大媒体任务内存并消除准入惊群

将同步与异步非文本任务的准入唤醒改为按任务和全局队首推进，批量续租等待者，避免大量请求同时争抢 advisory lock 和数据库连接。

对 Base64 请求解析、素材解码、上游媒体执行和结果物化增加分层并发限制，复用请求素材并释放重复 Gemini wire 数据；生产 API 默认入口解析并发 16，媒体物化并发 8。

新增 1000 个同步 Gemini 图像编辑请求的模拟上游压力验收，10 MiB 输入和输出下全部成功，Heap 峰值增长约 2.58 GiB，并验证共享上传哈希、单次 attempt 与本地零落盘。

验证：go test ./... -count=1；go vet ./...；gofmt -l 无输出；kubectl kustomize deploy/kubernetes/production；10 MiB Gemini Base64 千任务压力测试通过。

apps/api/internal/config/config.go

@@ -59,6 +59,8 @@ type Config struct {
 	BillingEngineMode                             string
 	ProcessRole                                   string
 	DatabaseMaxConns                              int
+	MediaRequestConcurrency                       int
+	MediaMaterializationConcurrency               int
 	AsyncQueueWorkerEnabled                       bool
 	AsyncWorkerHardLimit                          int
 	AsyncWorkerRefreshIntervalSeconds             int
@@ -117,6 +119,8 @@ func Load() Config {
 		BillingEngineMode:                 strings.ToLower(env("BILLING_ENGINE_MODE", "observe")),
 		ProcessRole:                       strings.ToLower(strings.TrimSpace(env("AI_GATEWAY_PROCESS_ROLE", "all"))),
 		DatabaseMaxConns:                  envInt("AI_GATEWAY_DATABASE_MAX_CONNS", 0),
+		MediaRequestConcurrency:           envIntValidated("AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY", 16),
+		MediaMaterializationConcurrency:   envIntValidated("AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY", 8),
 		AsyncQueueWorkerEnabled:           env("AI_GATEWAY_ASYNC_QUEUE_WORKER_ENABLED", "true") == "true",
 		AsyncWorkerHardLimit:              envIntValidated("AI_GATEWAY_ASYNC_WORKER_HARD_LIMIT", 2048),
 		AsyncWorkerRefreshIntervalSeconds: envIntValidated("AI_GATEWAY_ASYNC_WORKER_REFRESH_INTERVAL_SECONDS", 5),
@@ -132,6 +136,12 @@ func (c Config) Validate() error {
 	if c.DatabaseMaxConns < 0 || c.DatabaseMaxConns > 1000 {
 		return errors.New("AI_GATEWAY_DATABASE_MAX_CONNS must be between 1 and 1000 when configured")
 	}
+	if c.MediaMaterializationConcurrency != 0 && (c.MediaMaterializationConcurrency < 1 || c.MediaMaterializationConcurrency > 256) {
+		return errors.New("AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY must be between 1 and 256")
+	}
+	if c.MediaRequestConcurrency != 0 && (c.MediaRequestConcurrency < 1 || c.MediaRequestConcurrency > 1024) {
+		return errors.New("AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY must be between 1 and 1024")
+	}
 	switch strings.ToLower(strings.TrimSpace(c.BillingEngineMode)) {
 	case "", "observe", "enforce", "hold":
 	default:

apps/api/internal/config/config_test.go

@@ -131,6 +131,24 @@ func TestValidateProcessRoleAndDatabasePool(t *testing.T) {
 	if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "DATABASE_MAX_CONNS") {
 		t.Fatalf("Validate() error = %v, want invalid database max conns", err)
 	}
+	cfg.DatabaseMaxConns = 16
+	cfg.MediaMaterializationConcurrency = 257
+	if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "MEDIA_MATERIALIZATION_CONCURRENCY") {
+		t.Fatalf("Validate() error = %v, want invalid media materialization concurrency", err)
+	}
+	t.Setenv("AI_GATEWAY_MEDIA_MATERIALIZATION_CONCURRENCY", "12")
+	if got := Load().MediaMaterializationConcurrency; got != 12 {
+		t.Fatalf("media materialization concurrency = %d, want 12", got)
+	}
+	cfg = Load()
+	cfg.MediaRequestConcurrency = 1025
+	if err := cfg.Validate(); err == nil || !strings.Contains(err.Error(), "MEDIA_REQUEST_CONCURRENCY") {
+		t.Fatalf("Validate() error = %v, want invalid media request concurrency", err)
+	}
+	t.Setenv("AI_GATEWAY_MEDIA_REQUEST_CONCURRENCY", "96")
+	if got := Load().MediaRequestConcurrency; got != 96 {
+		t.Fatalf("media request concurrency = %d, want 96", got)
+	}
 }
 
 func TestValidateTaskHistorySettings(t *testing.T) {