Add benchmarks for node only mode and mock timers.

[luanmuniz]

Adds another small set of node:test benchmarks for areas listed in #55723.
A continuation of the work I started in #63754
This is a different PR because it brings a different set of attention points, and to keep both PR easy to review.

This PR covers:

• only mode with --test-only
• mock.timers.enable()
• mocked setTimeout
• mocked setInterval
• mocked setImmediate
• mocked scheduler.wait
• mocked AbortSignal.timeout
• mocked Date.now()
• mock.timers.setTime()
• mock.timers.runAll()

Notes

1. enable-* test the enable() + reset() instead of just enable. This happens because an error is thrown if enable is called multiple times without reset.
2. The scheduler.wait mode includes the final Promise.all() inside the measured section. My evaluation was that it was more appropriate to do so than not, since the scheduled waits are promise-based, but I am happy to adjust the structure if reviewers prefer a narrower measure.

Issue found while writing this

TLDR: I found an issue (that doesn't affect this PR) with using setImmediate and runAll together while writing the benchmark; guidance welcome;

While writing the benchmark, I found that setImmediate doesn't work with runAll(). The mocked setImmediate is scheduled with a special -1 time so it runs before zero-delay timers. runAll() computes longestTimer.runAt - this.#now in mock_timers.js:816, which becomes -1, and tick() rejects negative time in mock_timers.js:556.

I would love to open a new PR to fix this issue, but this is beyond my context of how the team working on this operates and what they think would be an appropriate fix. If the person reviewing the PR could provide some guidance on the best way to handle this issue, it would be much appreciated.

This issue doesn't impact this PR, all benchmarks still work with the current code (With tick(0) instead of runAll() for setImmediate).

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[RafaelGSS]

This is unnecessary; it's a clojure. Adding an assert.ok is probably more useful.

[luanmuniz]

Hi @RafaelGSS, thanks for taking a look.

Just to make sure I understand the suggestion: are you suggesting replacing the avoidV8Optimization assignment with an assertion inside the benchmarked section?

I added the avoidV8Optimization bit to avoid making the benchmarked callback completely empty, as detailed in the previous PR: #63754 (comment)

My concern with adding assert.ok() inside the callback is that the assertion itself would become part of what is being measured. Adding extra work inside the benchmark was also something @avivkeller raised in the previous PR: #63754 (comment)

I’m happy to adjust it either way. Please let me know how to proceed.

Thank you for the guidance.

Obs: I see the comments about calbrate-n on the first PR and i will push that change very soon here too

[luanmuniz]

Hi there, sorry for the force-push. Like the other branch, I had a local git history issue while cleaning up the branch, and I force-pushed to get the PR back into the intended state. Apologies for the noise. I’ll be more careful with the branch history going forward.

I've used calibrate-n, and there are the results in case you want to take a look:
calibrate-mock-timers.txt
calibrate-test-only.txt

The numbers are already adjusted.

cc @RafaelGSS