You're 100% right though. We don't have a fully baked story yet for application level testing. It's something we're trying to push forward as soon as we can.
From product adoption perspective, you guys are going to win a lot more users if you can support the following scenario:
1) unit-tests
2) integration-tests
3) provide at least guidance/best-practices on client-side testing in the _headless_ scenario (no complex setup of remote machines with 3-4 browsers installed etc)
People may have their opinions about testing UI/client-side (is it end-to-end? is it worth? not testing using the browser no-go?, etc) but the fact that Meteor is leaning toward more code in the client-side will definitely put testing in almost make-or-break decision for a group of developers.
GWT with their MVP approach is definitely heading toward that direction and it is quite unfortunate that the client-side JS community out there haven't picked up that style (some brushed GWT off because it is Java).
Rails won the heart of many Java developers not only because of its simplicity but also because of automated testing.
So, show how easy it is to test the whole Meteor app and I (almost) guarantee you will get many more developers (especially those who skipped the Rails boat).
https://github.com/meteor/meteor/tree/master/packages/
You're 100% right though. We don't have a fully baked story yet for application level testing. It's something we're trying to push forward as soon as we can.