Do not fail on close() if test Kudu server is not initialized

[losipiuk]

no release notes/docs

[findepi]

Note: I'm fine with the change, but i am also fine with afterclass failing if beforeclass failed.
Perhaps it helps read the logs?

[hashhar]

Is Kudu server not being initialized when after-class is called an expected situation? Are we masking some existing issue?

[losipiuk]

Is Kudu server not being initialized when after-class is called an expected situation? Are we masking some existing issue?

With alwaysRun it is always called. Even if @BeforeClass method which is supposed to set the kuduServer fails before actually doing so.
It is not masking anything I think. We still get failure reports from intialization.

[losipiuk]

Note: I'm fine with the change, but i am also fine with afterclass failing if beforeclass failed.
Perhaps it helps read the logs?

AfterClass failure does not provide any useful information. It just adds noise into the logs. The actual failure is in tests initialization.