[core] Report performance test results on each PR

[m4theushw]

This PR removes the performance tests that were running as unit test. Instead, they run now in a different environment, with the React production build and closer to how the user interacts with the grid. The results will be reported as a comment in each PR, that means that the PR, initially, won't fail if a regression is introduced.

The comparison between these results and a "gold master" (assumed to be the last released version) will be done after the merge and once we have a performance snapshot for a release.

As inspiration: https://github.com/saucelabs/performance-CI-demo/blob/main/README.md

[github-actions]

This pull request has conflicts, please resolve those before we can evaluate the pull request.

[oliviertassinari]

Could you set DANGER_GITHUB_API_TOKEN on this repo?

@m4theushw Done, added to CircleCI. The comment should come from https://github.com/mui-bot going forward. I have followed https://danger.systems/guides/getting_started.html

[mui-bot]

These are the results for the performance tests:

Test case	Unit	Min	Max	Median	Mean	σ
Filter 100k rows	ms	239.7	375.2	311.6	303	46.9
Sort 100k rows	ms	472.2	794.8	718	672.74	125.891
Select 100k rows	ms	217.3	315.5	268	267.28	32.912
Deselect 100k rows	ms	99.7	277.9	231.1	204.3	58.898

Generated by 🚫 dangerJS against c44a33d

[flaviendelangle]

@m4theushw could we add the standard deviation or the variance to have a glance of the value accurateness ?
I will not treat the same way a variation à 10ms if the variance is of 1ms or 50ms.

[m4theushw]

@flaviendelangle I added the standard deviation. As you can see, the values vary substantially, probably because integration between Playwright and the Chrome DevTools protocol is slow. This means that we won't be able to see if a PR has caused a small impact on performance. However, it's going to be useful for situations where a change causes a drastic performance hit, in which a unit test, using a small dataset, wouldn't fail due timeout. In 154cc7e I intentionally caused a performance degradation (ignore the failing unit tests). The average time to select all rows is 60x higher. This is a clear signal that something is wrong.

I'll revert 154cc7e once you react to this comment.

[flaviendelangle]

Seems perfect 👍