TxTimeoutTestCase#test_negativeTxTimeoutTest fails intermittently

Example is here: https://ci.wildfly.org/viewLog.html?buildId=88880&buildTypeId=WFPR

Looking at the code I think this test is no longer valid. Originally when the test was added the design was that the reaper would never close the EM, however it looks like this was updated in WFLY-4923 so the new requirement is that the last thread to be associated with the EM is responsible for closing. If the reaper associates, then the app dissociates, then the reaper dissociates then I think it is correct and expected for the reaper thread to close the EM.

As such I will submit a PR to delete this test.