Mii app patching test

[doxiao]

1. Add test case Copy edits to fix minor typos #5 and Removed use of italics for highlighting key terms #6, which patches an existing app or adds an additional app respectively, including verifications of patching, app accessibility, and app continuous availability.
2. Use Kubernetes client exec api to verify app availability.
3. Add timeout to the join call in Kubernetes API exec command action.
4. Cleanup, including removing ItWitValidation.java because its content has been moved into ItMiiDomain.java.

[doxiao]

Clean test run on external Jenkins - https://build.weblogick8s.org:8443/job/weblogic-kubernetes-operator-model-in-image-tests-10/23/

[doxiao]

The current version is clean in external Jenkins run https://build.weblogick8s.org:8443/job/weblogic-kubernetes-operator-model-in-image-tests-10/37/.

[vanajamukkara]

I think the new tests in this PR should be added in a new or other existing It class where the domain creation/existence is a prerequisite, then you don't need to order the tests. Here it is assumed that the domain is up and running by ordering the tests where it should have been verified in beforeEach method. This is a late comment in this PR. If everyone agrees, changes can be made in another PR.

[doxiao]

Thanks, and yes, I would prefer address this comment in a different PR.

I still need to understand how to maintain expected order among different test methods without using @order annotation, or between different test classes.

[doxiao]

In order for the patching test cases to be fully independent, we need to do three things:

1. Move those test methods into a different test class;
2. Create a new initial domain in the BeforeAll handler of the new class; instead of using the domain that is created in the first end-to-end test case;
3. At the end of each test case, rollback what we have done in the method so that the domain goes back to the initial state, to make the state predictable for the next test case.

Both 2) and 3) will increase the total time of the test suite run, which is what the current approach tries to avoid.

[markxnelson]

there are still some comments that have not been adequately addressed. my biggest concern is the use of exec() to run kubectl exec. @rjeberhard i approved this, in the interests of getting the good stuff here into the regular test runs, i am willing to overlook some of the not ideal stuff, if you are. i will let you decide whether to merge this as-is, or persist with those last few issues