test: fix flaky test-performance-eventloopdelay

The test was flaky because the histogram sampling might not have
completed before the assertions ran. By using setImmediate before
disabling the histogram and checking values, we give the event loop
a chance to record final samples on slower systems.
Also added an explicit check for histogram.count > 0 to provide a
clearer error message if no samples are recorded.
Refs: nodejs/reliability#1450
PR-URL: #61629
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>

Author: mcollina

test/sequential/test-performance-eventloopdelay.js

@@ -64,42 +64,47 @@ const { sleep } = require('internal/util');
 if(--m>0){
 setTimeout(spinAWhile,common.platformTimeout(500));
 }else{
-histogram.disable();
-// The values are non-deterministic, so we just check that a value is
-// present, as opposed to a specific value.
-assert(histogram.min>0);
-assert(histogram.max>0);
-assert(histogram.stddev>0);
-assert(histogram.mean>0);
-assert(histogram.percentiles.size>0);
-for(letn=1;n<100;n=n+0.1){
-assert(histogram.percentile(n)>=0);
-}
-histogram.reset();
-assert.strictEqual(histogram.min,9223372036854776000);
-assert.strictEqual(histogram.max,0);
-assert(Number.isNaN(histogram.stddev));
-assert(Number.isNaN(histogram.mean));
-assert.strictEqual(histogram.percentiles.size,1);
+// Give the histogram a chance to record final samples before disabling.
+// This helps on slower systems where sampling may be delayed.
+setImmediate(common.mustCall(()=>{
+histogram.disable();
+// The values are non-deterministic, so we just check that a value is
+// present, as opposed to a specific value.
+assert(histogram.count>0,`Expected samples to be recorded, got count=${histogram.count}`);
+assert(histogram.min>0);
+assert(histogram.max>0);
+assert(histogram.stddev>0);
+assert(histogram.mean>0);
+assert(histogram.percentiles.size>0);
+for(letn=1;n<100;n=n+0.1){
+assert(histogram.percentile(n)>=0);
+}
+histogram.reset();
+assert.strictEqual(histogram.min,9223372036854776000);
+assert.strictEqual(histogram.max,0);
+assert(Number.isNaN(histogram.stddev));
+assert(Number.isNaN(histogram.mean));
+assert.strictEqual(histogram.percentiles.size,1);
 
-['a',false,{},[]].forEach((i)=>{
-assert.throws(
-()=>histogram.percentile(i),
-{
-name: 'TypeError',
-code: 'ERR_INVALID_ARG_TYPE'
-}
-);
-});
-[-1,0,101,NaN].forEach((i)=>{
-assert.throws(
-()=>histogram.percentile(i),
-{
-name: 'RangeError',
-code: 'ERR_OUT_OF_RANGE'
-}
-);
-});
+['a',false,{},[]].forEach((i)=>{
+assert.throws(
+()=>histogram.percentile(i),
+{
+name: 'TypeError',
+code: 'ERR_INVALID_ARG_TYPE',
+}
+);
+});
+[-1,0,101,NaN].forEach((i)=>{
+assert.throws(
+()=>histogram.percentile(i),
+{
+name: 'RangeError',
+code: 'ERR_OUT_OF_RANGE',
+}
+);
+});
+}));
 }
 }
 spinAWhile();