Fix load test scenarios, runner, and CI threshold checks

- Add constant-arrival-rate scenarios to all 4 service scripts (api,
  darkwatch, spamshield, voiceprint) to enforce 500 req/s target
- Fix defaultThresholds() to return { thresholds: {...} } so
  http_req_duration and errors thresholds are actually applied
- Rewrite run-all.sh: per-service summary files, proper env var
  passing (DURATION, API_TOKEN), fixed threshold aggregation
- Update CI workflow threshold check jq to match new threshold-results
  structure (.services.<name>.exitCode)

Co-Authored-By: Paperclip <noreply@paperclip.ing>

scripts/load-test/services/api.js

@@ -7,7 +7,22 @@ const errorRate = new Rate('errors');
 const notificationLatency = new Trend('notification_p99');
 const correlationLatency = new Trend('correlation_p99');
 
+const TARGET_RPS = getTargetRps();
+const DURATION = getDuration();
+
 export const options = {
+  scenarios: {
+    sustained_load: {
+      executor: 'constant-arrival-rate',
+      duration: DURATION,
+      rate: TARGET_RPS,
+      preAllocatedVUs: 20,
+      maxVUs: 100,
+      startTime: '0s',
+      exec: 'default',
+      tags: { scenario: 'sustained_load' },
+    },
+  },
   thresholds: {
     ...defaultThresholds(250).thresholds,
     notification_p99: ['p(99)<500'],

scripts/load-test/services/darkwatch.js

@@ -8,7 +8,22 @@ const scanLatency = new Trend('scan_p99');
 const watchlistLatency = new Trend('watchlist_p99');
 const alertLatency = new Trend('alert_p99');
 
+const TARGET_RPS = getTargetRps();
+const DURATION = getDuration();
+
 export const options = {
+  scenarios: {
+    sustained_load: {
+      executor: 'constant-arrival-rate',
+      duration: DURATION,
+      rate: TARGET_RPS,
+      preAllocatedVUs: 20,
+      maxVUs: 100,
+      startTime: '0s',
+      exec: 'default',
+      tags: { scenario: 'sustained_load' },
+    },
+  },
   thresholds: {
     ...defaultThresholds(200).thresholds,
     scan_p99: ['p(99)<300'],

scripts/load-test/services/spamshield.js

@@ -8,7 +8,22 @@ const smsClassifyP99 = new Trend('sms_classify_p99');
 const numberReputationP99 = new Trend('number_reputation_p99');
 const callAnalyzeP99 = new Trend('call_analyze_p99');
 
+const TARGET_RPS = getTargetRps();
+const DURATION = getDuration();
+
 export const options = {
+  scenarios: {
+    sustained_load: {
+      executor: 'constant-arrival-rate',
+      duration: DURATION,
+      rate: TARGET_RPS,
+      preAllocatedVUs: 20,
+      maxVUs: 100,
+      startTime: '0s',
+      exec: 'default',
+      tags: { scenario: 'sustained_load' },
+    },
+  },
   thresholds: {
     ...defaultThresholds(400).thresholds,
     sms_classify_p99: ['p(99)<150'],

scripts/load-test/services/voiceprint.js

@@ -8,7 +8,22 @@ const enrollmentLatency = new Trend('enrollment_p99');
 const verificationLatency = new Trend('verification_p99');
 const modelLatency = new Trend('model_retrieval_p99');
 
+const TARGET_RPS = getTargetRps();
+const DURATION = getDuration();
+
 export const options = {
+  scenarios: {
+    sustained_load: {
+      executor: 'constant-arrival-rate',
+      duration: DURATION,
+      rate: TARGET_RPS,
+      preAllocatedVUs: 20,
+      maxVUs: 100,
+      startTime: '0s',
+      exec: 'default',
+      tags: { scenario: 'sustained_load' },
+    },
+  },
   thresholds: {
     ...defaultThresholds(250).thresholds,
     enrollment_p99: ['p(99)<500'],