feat(queue): 增加非文本模型分布式准入队列

使用 PostgreSQL 统一同步与异步非文本任务的并发准入、持久化等待和 Worker 容量分配，并将生产 API 与独立 Worker 角色拆分。

补充策略管理、共享契约、OpenAPI、Kubernetes 双节点 Worker 清单及跨节点验收脚本；未默认启用任何生产 queue_size 策略。

已在原基线完成 Go、前端、迁移、Shell、Kustomize 与长任务容量验收；合入最新主干后将重新执行发布门禁。

apps/api/internal/store/rate_limit_policy_test.go

@@ -2,6 +2,66 @@ package store
 
 import "testing"
 
+func TestNormalizeAndValidateQueuePolicy(t *testing.T) {
+	t.Run("defaults maximum wait and preserves extensions", func(t *testing.T) {
+		policy, err := NormalizeAndValidateRateLimitPolicy(map[string]any{
+			"strategy": "strict",
+			"rules": []any{map[string]any{
+				"metric": "queue_size",
+				"limit":  12,
+				"source": "admin",
+			}},
+		})
+		if err != nil {
+			t.Fatalf("NormalizeAndValidateRateLimitPolicy() error = %v", err)
+		}
+		queue, ok := QueueRuleFromPolicy(policy)
+		if !ok || queue.Limit != 12 || queue.MaxWaitSeconds != 600 {
+			t.Fatalf("queue rule = %+v, %v", queue, ok)
+		}
+		rules := policy["rules"].([]any)
+		rule := rules[0].(map[string]any)
+		if rule["source"] != "admin" {
+			t.Fatalf("unknown extension was lost: %+v", rule)
+		}
+	})
+
+	t.Run("normalizes zero to a removed queue rule", func(t *testing.T) {
+		policy, err := NormalizeAndValidateRateLimitPolicy(map[string]any{
+			"rules": []any{
+				map[string]any{"metric": "rpm", "limit": 30},
+				map[string]any{"metric": "queue_size", "limit": 0},
+			},
+		})
+		if err != nil {
+			t.Fatalf("NormalizeAndValidateRateLimitPolicy() error = %v", err)
+		}
+		if _, ok := QueueRuleFromPolicy(policy); ok {
+			t.Fatalf("zero queue rule remained enabled: %+v", policy)
+		}
+		rules := policy["rules"].([]any)
+		if len(rules) != 1 {
+			t.Fatalf("rules length = %d, want 1", len(rules))
+		}
+	})
+
+	for _, test := range []struct {
+		name   string
+		policy map[string]any
+	}{
+		{name: "fractional queue", policy: map[string]any{"rules": []any{map[string]any{"metric": "queue_size", "limit": 1.5}}}},
+		{name: "queue too large", policy: map[string]any{"rules": []any{map[string]any{"metric": "queue_size", "limit": 10001}}}},
+		{name: "wait too large", policy: map[string]any{"rules": []any{map[string]any{"metric": "queue_size", "limit": 1, "maxWaitSeconds": 3601}}}},
+		{name: "wait on rpm", policy: map[string]any{"rules": []any{map[string]any{"metric": "rpm", "limit": 1, "maxWaitSeconds": 60}}}},
+	} {
+		t.Run(test.name, func(t *testing.T) {
+			if _, err := NormalizeAndValidateRateLimitPolicy(test.policy); err == nil {
+				t.Fatalf("invalid policy was accepted: %+v", test.policy)
+			}
+		})
+	}
+}
+
 func TestEffectiveRateLimitPolicyPrecedence(t *testing.T) {
 	policy := func(limit float64) map[string]any {
 		return map[string]any{"rules": []any{map[string]any{"metric": "concurrent", "limit": limit}}}