timers: fix unsafe array iteration

[RaisinTen]

Fixes: #37222

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/35937/

[RaisinTen]

Benchmark CI for timers: https://ci.nodejs.org/view/Node.js%20benchmark/job/benchmark-node-micro-benchmarks/930/
Using SafeArrayIterator has resulted in a significant perf regression:

Benchmark results

19:56:10 timers/immediate.jstype='breadth4' n=5000000 * -4.65 % ±3.97% ±5.30% ±6.92%
19:56:10 timers/immediate.jstype='depth1' n=5000000 * 1.93 % ±1.86% ±2.49% ±3.26%
19:56:10 timers/timers-breadth-args.jsn=1000000 *** -24.02 % ±5.82% ±7.78% ±10.19%

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/35940/

[RaisinTen]

New Benchmark CI (using ReflectApply):
https://ci.nodejs.org/view/Node.js%20benchmark/job/benchmark-node-micro-benchmarks/932/

Benchmark results

 confidence improvement accuracy (*) (**) (***)
timers/immediate.jstype='breadth1' n=5000000 0.50 % ±3.18% ±4.23% ±5.51%
timers/immediate.jstype='breadth4' n=5000000 -1.48 % ±4.55% ±6.06% ±7.89%
timers/immediate.jstype='breadth' n=5000000 -1.31 % ±3.90% ±5.18% ±6.75%
timers/immediate.jstype='clear' n=5000000 1.78 % ±3.34% ±4.45% ±5.80%
timers/immediate.jstype='depth1' n=5000000 1.06 % ±2.53% ±3.37% ±4.40%
timers/immediate.jstype='depth' n=5000000 -0.27 % ±2.51% ±3.35% ±4.37%
timers/set-immediate-breadth-args.jsn=5000000 -2.31 % ±3.44% ±4.57% ±5.95%
timers/set-immediate-breadth.jsn=10000000 -1.88 % ±4.10% ±5.46% ±7.11%
timers/set-immediate-depth-args.jsn=5000000 0.42 % ±1.57% ±2.09% ±2.72%
timers/timers-breadth-args.jsn=1000000 1.50 % ±6.38% ±8.48% ±11.04%
timers/timers-breadth.jsn=5000000 1.18 % ±3.17% ±4.22% ±5.50%
timers/timers-cancel-pooled.jsn=5000000 8.24 % ±11.36% ±15.12% ±19.69%
timers/timers-cancel-unpooled.jsdirection='end' n=1000000 -1.42 % ±8.14% ±10.83% ±14.10%
timers/timers-cancel-unpooled.jsdirection='start' n=1000000 -2.38 % ±5.38% ±7.16% ±9.32%
timers/timers-depth.jsn=1000 0.48 % ±0.78% ±1.04% ±1.35%
timers/timers-insert-pooled.jsn=5000000 0.65 % ±3.05% ±4.06% ±5.29%
timers/timers-insert-unpooled.jsdirection='end' n=1000000 -2.16 % ±3.85% ±5.12% ±6.66%
timers/timers-insert-unpooled.jsdirection='start' n=1000000 1.73 % ±5.02% ±6.68% ±8.70%
timers/timers-timeout-nexttick.jsn=50000 -0.24 % ±2.17% ±2.89% ±3.76%
timers/timers-timeout-nexttick.jsn=5000000 0.25 % ±6.66% ±8.87% ±11.55%
timers/timers-timeout-pooled.jsn=10000000 -0.62 % ±6.87% ±9.15% ±11.91%
timers/timers-timeout-unpooled.jsn=1000000 * -12.70 % ±9.87% ±13.19% ±17.28%
Be aware that when doing many comparisons the risk of a false-positive
result increases. In this case, there are 22 comparisons, you can thus
expect the following amount of false-positive results:
1.10 false positives, when considering a 5% risk acceptance (*, **, ***),
0.22 false positives, when considering a 1% risk acceptance (**, ***),
0.02 false positives, when considering a 0.1% risk acceptance (***)

[aduh95]

Benchmark results are a bit weird, I kicked off a new benchmark CI to see if it's just a flakiness.
https://ci.nodejs.org/view/Node.js%20benchmark/job/benchmark-node-micro-benchmarks/934/

[Fishrock123]

I think Reflect probably causes more overhead than just doing a more manual check.

[aduh95]

Benchmark CI results look good 🎉

Details

 confidence improvement accuracy (*) (**) (***)
timers/immediate.jstype='breadth1' n=5000000 1.00 % ±4.06% ±5.41% ±7.04%
timers/immediate.jstype='breadth4' n=5000000 -1.00 % ±4.15% ±5.53% ±7.21%
timers/immediate.jstype='breadth' n=5000000 3.12 % ±4.05% ±5.38% ±7.01%
timers/immediate.jstype='clear' n=5000000 -0.04 % ±2.15% ±2.86% ±3.72%
timers/immediate.jstype='depth1' n=5000000 -0.03 % ±1.74% ±2.31% ±3.02%
timers/immediate.jstype='depth' n=5000000 0.75 % ±2.22% ±2.95% ±3.85%
timers/set-immediate-breadth-args.jsn=5000000 0.19 % ±4.62% ±6.16% ±8.03%
timers/set-immediate-breadth.jsn=10000000 * 4.00 % ±3.89% ±5.18% ±6.74%
timers/set-immediate-depth-args.jsn=5000000 * 2.03 % ±1.80% ±2.40% ±3.14%
timers/timers-breadth-args.jsn=1000000 -5.37 % ±5.51% ±7.33% ±9.54%
timers/timers-breadth.jsn=5000000 0.81 % ±3.38% ±4.50% ±5.86%
timers/timers-cancel-pooled.jsn=5000000 1.91 % ±10.69% ±14.22% ±18.51%
timers/timers-cancel-unpooled.jsdirection='end' n=1000000 -0.92 % ±5.95% ±7.91% ±10.30%
timers/timers-cancel-unpooled.jsdirection='start' n=1000000 0.18 % ±6.67% ±8.89% ±11.58%
timers/timers-depth.jsn=1000 -0.13 % ±0.65% ±0.87% ±1.14%
timers/timers-insert-pooled.jsn=5000000 -0.54 % ±2.80% ±3.74% ±4.88%
timers/timers-insert-unpooled.jsdirection='end' n=1000000 1.44 % ±4.47% ±5.95% ±7.74%
timers/timers-insert-unpooled.jsdirection='start' n=1000000 0.55 % ±5.13% ±6.82% ±8.88%
timers/timers-timeout-nexttick.jsn=50000 0.14 % ±2.20% ±2.93% ±3.83%
timers/timers-timeout-nexttick.jsn=5000000 0.13 % ±3.60% ±4.80% ±6.25%
timers/timers-timeout-pooled.jsn=10000000 * 8.82 % ±8.47% ±11.29% ±14.72%
timers/timers-timeout-unpooled.jsn=1000000 0.62 % ±13.21% ±17.58% ±22.88%
Be aware that when doing many comparisons the risk of a false-positive
result increases. In this case, there are 22 comparisons, you can thus
expect the following amount of false-positive results:
1.10 false positives, when considering a 5% risk acceptance (*, **, ***),
0.22 false positives, when considering a 1% risk acceptance (**, ***),
0.02 false positives, when considering a 0.1% risk acceptance (***)

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/35949/

[benjamingr]

Ping @nodejs/timers

[Trott]

Landed in c0e66e3