[hotfix]Intermittent unit test failures we have hit in ExecutionGraph tests

[davidradl]

We were getting an intermittent error in our CI pipeline, as below. This change adds a small sleep to allow the 2 slow tasks to start before it checks they are there. This approach is in line with other parts of this code e.g. this

[ERROR] Tests run: 13, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 2.068 s <<< FAILURE! -- in org.apache.flink.runtime.scheduler.slowtaskdetector.ExecutionTimeBasedSlowTaskDetectorTest
#24 511.0 [ERROR] org.apache.flink.runtime.scheduler.slowtaskdetector.ExecutionTimeBasedSlowTaskDetectorTest.testBalancedInput -- Time elapsed: 0.011 s <<< FAILURE!
#24 511.0 java.lang.AssertionError:
#24 511.0
#24 511.0 Expected size: 2 but was: 0 in:
#24 511.0 {}
#24 511.0 at org.apache.flink.runtime.scheduler.slowtaskdetector.ExecutionTimeBasedSlowTaskDetectorTest.testBalancedInput(ExecutionTimeBasedSlowTaskDetectorTest.java:269)
#24 511.0 at java.base/java.lang.reflect.Method.invoke(Method.java:586)
#24 511.0 at java.base/java.util.concurrent.ForkJoinTask.doExec(ForkJoinTask.java:387)
#24 511.0 at java.base/java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(ForkJoinPool.java:1312)
#24 511.0 at java.base/java.util.concurrent.ForkJoinPool.scan(ForkJoinPool.java:1843)
#24 511.0 at java.base/java.util.concurrent.ForkJoinPool.runWorker(ForkJoinPool.java:1808)
#24 511.0 at java.base/java.util.concurrent.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:188)

[flinkbot]

CI report:

• 2613629 Azure: FAILURE

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build

[davidradl]

We have hit 2 more intermittent unit test failures in this area. Fixing those here too.

[ERROR]   ExecutionTimeBasedSlowTaskDetectorTest.testNoFinishedTaskButRatioIsZero:81 
#24 546.3 Expected size: 3 but was: 0 i

and

ExecutionGraphRestartTest.testCancelWhileFailing:217 
#24 541.9 expected: RUNNING
#24 541.9  but was: FAILING

[davidradl]

also IBM Bob is adding .metals to github repos, so I have git ignored this folder.

[snuyanzin]

We should not rely on timeouts in tests
this is is also mentioned in contribution guide https://flink.apache.org/how-to-contribute/code-style-and-quality-common/#avoid-timeouts-in-junit-tests

[davidradl]

OK I was following the existing approach in the code. Also I see ExecutionGraphTestUtils.waitUntilJobStatus that is effectively a method doing the timeout. The comment says // this is a poor implementation - we may want to improve it eventually

I will investigate to see if there is another way,