I have plenty thorough testing practices for client-side code already; but there's also reasons to test that same code at the command-line:
- Coverage support. I can't get code-coverage out of the browser.
- Pre-C.I. and server-C.I. support. It'd be nice to have information about server- and client-side test status in one place; and this could be the case with Travis-CI if we PhantomJS-tested the client-side code.
To boot, there's little reason not to. More wide-spread testing is always better.