Stabilize the CI

[sidharthv96]

We're getting intermittent failures in CI, which reduces our ability to spot actual issues.

157c90e and c63ea3e had completely unrelated failures, which were fixed when rerun.

[sidharthv96]

I'm uncertain if merging next would fix some of these, as the build pipeline changes to esbuild.

cc @aloisklink @nirname @Yokozuna59

[nirname]

We need more examples when we are talking about unstable CI. This very example that you have provided is related to gantt optimization specifically, I presume. I can reproduce this failure locally, without CI. It does not work in Docker environment. May be I'll give it a shot and track down the problem.

If there are more tests scenarios which produce unpredictable results in CI, that worth investigating. We need like a critical mass of examples to figure out what is going on. Too many different potential roots to judge by one occasion.

One of the symptoms of "unstable CI" is that rerunning action helps to fix it, as you have already mentioned. I'll try to search for more representative cases, apart from those two.

[sidharthv96]

Nope, both the failures have nothing to do with Gantt.
They were two different tests in flowchart, failing due to two different reasons (both seemed to be build related)

The Gantt failure was not a flaky failure, but transient failures like this have a part in why that issue was overlooked.