When working on WFLY-19755 I dug into the various appclient related DeploymentUnitProcessors, and I can't see the need for ApplicationClientStructureProcessor. Perhaps it was needed before the WFLY-19755 fix, although that's not obviously the case. It's only registered if the user specifies a particular deployment and it only deals with ears – processing the top level deployment and the subdeployment. The test added for WFLY-19755 demonstrates specifying a particular ear subdeployment isn't required. ApplicationClientStructureProcessor doesn't do anything special; everything it does would need to be done elsewhere if this DUP isn't registered. The WFLY-19755 test demonstrates that what it does is getting done elsewhere, so this DUP seems like cruft now.

Any fix for this should not be merged close at all before any release tag, as running the TCK is the main way we test appclient. We don't want to disrupt TCK runs anywhere near a release tag.