perf(queue): 按策略动态扩缩异步 Worker

将 River 执行容量按平台模型与用户组的有效并发策略动态调整，并为长任务续租、并发租约原子抢占和限流退避补充保护。\n\n统一平台模型限流继承语义，兼容历史 platformLimits/modelLimits，并为三个迁移图像模型建立独立并发租约。补充管理端显式继承/覆盖配置、指标、单元测试及隔离 PostgreSQL 验收。\n\n验证：go test ./... -count=1；pnpm lint；pnpm test；pnpm build；隔离 PostgreSQL 并发原子性/续租测试；128 任务与三分钟长任务动态 Worker 验收。

apps/api/migrations/0081_image_gateway_rate_limits.sql

@@ -0,0 +1,44 @@
+-- The three migrated image identities previously relied on a global
+-- user-group limit. Give every upstream binding its own canonical concurrency
+-- lease so raising the service group no longer removes provider protection.
+WITH limits(invocation_name, policy) AS (
+  VALUES
+    (
+      'gemini-3-pro-image',
+      '{"rules":[{"metric":"concurrent","limit":10,"leaseTtlSeconds":600}]}'::jsonb
+    ),
+    (
+      'gemini-3.1-flash-image',
+      '{"rules":[{"metric":"concurrent","limit":10,"leaseTtlSeconds":600}]}'::jsonb
+    ),
+    (
+      'gpt-image-2',
+      '{"rules":[{"metric":"concurrent","limit":5,"leaseTtlSeconds":300}]}'::jsonb
+    )
+),
+updated_base_models AS (
+  UPDATE base_model_catalog base_model
+  SET default_rate_limit_policy = limits.policy,
+      metadata = COALESCE(base_model.metadata, '{}'::jsonb) || jsonb_build_object(
+        'rateLimitSource', 'image-gateway-migration',
+        'rateLimitUpdatedAt', '2026-07-24'
+      ),
+      updated_at = now()
+  FROM limits
+  WHERE base_model.invocation_name = limits.invocation_name
+  RETURNING base_model.id
+)
+UPDATE platform_models platform_model
+SET rate_limit_policy = '{}'::jsonb,
+    rate_limit_policy_mode = 'inherit',
+    updated_at = now()
+FROM updated_base_models base_model
+WHERE platform_model.base_model_id = base_model.id
+  AND (
+    platform_model.rate_limit_policy = '{}'::jsonb
+    OR platform_model.rate_limit_policy = '{"rules":[]}'::jsonb
+    OR platform_model.rate_limit_policy ? 'platformLimits'
+    OR platform_model.rate_limit_policy ? 'modelLimits'
+    OR platform_model.rate_limit_policy ? 'platform_limits'
+    OR platform_model.rate_limit_policy ? 'model_limits'
+  );