Make BuildTriggerListener wait for onFinalized, not just onCompleted

[jglick]

As a result of jenkinsci/jenkins-test-harness#198 we see occasionally flakes in PCT

java.io.IOException: …/pipeline-build-step-plugin/target/tmp/j h17881095753960446783/jobs/dir1/jobs/downstream/builds/1/build.xml
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:144)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:131)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:131)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:131)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:131)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:131)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:131)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.dispose(TemporaryDirectoryAllocator.java:99)
	at org.jvnet.hudson.test.TestEnvironment.dispose(TestEnvironment.java:84)
	at org.jvnet.hudson.test.JenkinsRule.after(JenkinsRule.java:525)
	at org.jvnet.hudson.test.JenkinsRule$1.evaluate(JenkinsRule.java:626)
	at …
Caused by: java.nio.file.DirectoryNotEmptyException: …/pipeline-build-step-plugin/target/tmp/j h17881095753960446783/jobs/dir1/jobs/downstream/builds/1
	at java.base/sun.nio.fs.UnixFileSystemProvider.implDelete(UnixFileSystemProvider.java:247)
	at java.base/sun.nio.fs.AbstractFileSystemProvider.deleteIfExists(AbstractFileSystemProvider.java:110)
	at java.base/java.nio.file.Files.deleteIfExists(Files.java:1181)
	at org.jvnet.hudson.test.TemporaryDirectoryAllocator.delete(TemporaryDirectoryAllocator.java:141)
	... 14 more

with stderr like this:

…
1.030 [id=192]	INFO	jenkins.InitReactorRunner$1#onAttained: Completed initialization
1.145 [dir2/upstream #1] Started
1.145 [dir2/upstream #1] [Pipeline] Start of Pipeline
1.196 [dir1/downstream #1] Started by upstream project "dir2/upstream" build number 1
1.196 [dir1/downstream #1] Running as SYSTEM
1.196 [dir1/downstream #1] Building in workspace …/pipeline-build-step-plugin/target/tmp/j h17881095753960446783/workspace/dir1/downstream
1.196 [dir2/upstream #1] [Pipeline] build
1.196 [dir2/upstream #1] Scheduling project: dir1 » downstream
1.196 [dir2/upstream #1] Starting building: dir1 » downstream #1
1.246 [dir1/downstream #1] Sleeping 1ms
1.247 [dir2/upstream #1] Build dir1 » downstream #1 completed: SUCCESS
1.247 [dir2/upstream #1] [Pipeline] End of Pipeline
1.247 [dir2/upstream #1] Finished: SUCCESS
1.254 [id=169]	INFO	hudson.lifecycle.Lifecycle#onStatusUpdate: Stopping Jenkins
1.262 [id=169]	INFO	hudson.lifecycle.Lifecycle#onStatusUpdate: Jenkins stopped
1.312 [dir1/downstream #1] Finished: SUCCESS
1.311 [id=169]	INFO	o.j.h.t.TemporaryDirectoryAllocator#dispose: deleting …/pipeline-build-step-plugin/target/tmp/j h17881095753960446783

You can see that the downstream build is actually finishing after the test has completed and when JenkinsRule is trying to clean up. I cannot reproduce this locally but suspect that this is because the listener used by the upstream step is only waiting for the downstream build to be nearly complete. Probably best to wait for it to be fully complete, which should also prevent the test flake.

[jglick]

Might have also solved https://github.com/jenkinsci/pipeline-build-step-plugin/runs/15047321442 (which prevented release of #116, which might have solved unrelated flakes).