refactor(ops): 移除duration相关告警指标，简化监控配置

主要改动：
- 移除 p95_latency_ms 和 p99_latency_ms 告警指标类型
- 移除配置中的 latency_p99_ms_max 阈值设置
- 简化健康分数计算（移除latency权重，重新归一化SLA和错误率）
- 移除duration相关的诊断规则和阈值检查
- 统一术语：延迟 → 请求时长
- 保留duration数据展示，但不再用于告警判断
- 聚焦TTFT作为主要的响应速度告警指标

影响范围：
- Backend: handler, service, models, tests
- Frontend: API types, i18n, components

backend/internal/service/ops_health_score.go

@@ -32,7 +32,7 @@ func computeDashboardHealthScore(now time.Time, overview *OpsDashboardOverview)
 }
 
 // computeBusinessHealth calculates business health score (0-100)
-// Components: SLA (50%) + Error Rate (30%) + Latency (20%)
+// Components: SLA (50%) + Error Rate (30%)
 func computeBusinessHealth(overview *OpsDashboardOverview) float64 {
 	// SLA score: 99.5% → 100, 95% → 0 (linear)
 	slaScore := 100.0
@@ -59,22 +59,9 @@ func computeBusinessHealth(overview *OpsDashboardOverview) float64 {
 		}
 	}
 
-	// Latency score: 1s → 100, 10s → 0 (linear)
-	// Uses P99 of duration (TTFT is less critical for overall health)
-	latencyScore := 100.0
-	if overview.Duration.P99 != nil {
-		p99 := float64(*overview.Duration.P99)
-		if p99 > 1000 {
-			if p99 <= 10000 {
-				latencyScore = (10000 - p99) / 9000 * 100
-			} else {
-				latencyScore = 0
-			}
-		}
-	}
-
-	// Weighted combination
-	return slaScore*0.5 + errorScore*0.3 + latencyScore*0.2
+	// Weighted combination (renormalized after removing duration)
+	const weightSum = 0.8
+	return (slaScore*0.5 + errorScore*0.3) / weightSum
 }
 
 // computeInfraHealth calculates infrastructure health score (0-100)