RateLimitBranchPropertyTest.rateLimitsConcurrentBuilds is flaky too

[jglick]

Follows up #202. Flake noticed in jenkinsci/bom#301:

java.lang.AssertionError: 
At least the rate implied delay but no more than 500ms longer
Expected: (a value equal to or greater than <3400L> and a value less than or equal to <4100L>)
     but: a value equal to or greater than <3400L> <3342L> was less than <3400L>
	at org.hamcrest.MatcherAssert.assertThat(MatcherAssert.java:20)
	at jenkins.branch.RateLimitBranchPropertyTest.rateLimitsConcurrentBuilds(RateLimitBranchPropertyTest.java:292)

[bitwiseman]

We need to figure out how to reenable these rather than skipping when they fail.

[jglick]

how to reenable these

I guess. Looking at the test code, it is clearly inherently too flaky to run in CI.

[bitwiseman]

@jglick

Looking at the test code, it is clearly inherently too flaky to run in CI.

But without these tests the feature could break completely and we would not know.

[jglick]

But without these tests the feature could break completely and we would not know.

Indeed. We can only hope that in the unlikely event someone actually touches this code again, that they will run these tests locally to validate it. In the meantime, the test is causing problems for automation and sucking up developer time that could be better spent elsewhere.