The good part is that it's not that bad to fit within the company infrastructure, particularly if it's a BigCorp kind of thing. It's interesting because it allows you to play well with CI processes and quality rules - I'm thinking about tests here. With this kind of setup, you can make the front-end build fail if e.g. code coverage drops below a given threshold, using the same setup you use for other components.
The boring part is that it's yet another place to put config info. A particularly ugly one which reeks of over-engineering IMHO. Also it's XML; XML isn't quite the song of the front-end people, at least config-wise. :)
Saying that however to just build the client you can run "gulp" or "gulp zip" to create a build artefact.
Or for test execution "gulp run" then access http://localhost:4444/test
Anyone wanting to use this example would probably be better creating a "gulp test" task also.