Waiting for the JDT index

[LorenzoBettini]

Investigating on #3057, as suggested here https://download.eclipse.org/releases/2024-06/, I tried to wait for JDT index in some places where tests were flaky. I introduced a wait method here, with some logging, LorenzoBettini@63f6513#diff-980df7992443dd6c25d5264da62b20f4f63f831fa76f17246f702cc386da8d1aR421 and I added the wait for index mainly in Xtend IDE tests:

• LorenzoBettini@63f6513#diff-19ed180733058938a61d0d546fc9982715aa398e12c2e652680306e3bffa6415R198
• LorenzoBettini@63f6513#diff-53fafbe9aacbb66aebfec1beb8c16833a0172c55ecbc324fdc693f461958f84dR292
• LorenzoBettini@63f6513#diff-466b05d7c9b5a349cd23ab85040c3ad3da2ad29be8dc2c81a53e485f9341371aR58

flakes haven't disappeared in Windows and macOS but they reduced a lot and when there's a flake, on the second run, it succeeds.

I also added it here LorenzoBettini@63f6513#diff-22769996fc085327d4c4e67c51fe171bbc3db46c53330acb2b7b90214a545c1aR72 which was also flaky and in this case for the moment the flake disappeared. In this latter case, I don't think we use the JDT index, though. Maybe that's still helpful in warming up Eclipse?

This is the corresponding macOS job https://github.com/LorenzoBettini/xtext/actions/runs/9207206320/job/25329823688 with minimal flakes.

In general, when we use the JDT index in the code base shouldn't we wait for it to complete before using it?

I mean here

https://github.com/eclipse/xtext/blob/1b97ae8afaa6eefd9c10f5f3c8591d5e8eb2e00b/org.eclipse.xtext.common.types.ui/src/org/eclipse/xtext/common/types/access/jdt/JdtTypeProvider.java#L368

I also wonder whether in tests we should always wait for it before issuing a build and after; i.e., within waitForBuild.

cc @cdietrich @szarnekow