
We got caught in the middle of this policy change when we uploaded an update last month and it ran into a problem with 3.0. We immediately corrected that problem and tested our app on 3.0 on our own devices and at the Apple lab in Cupertino. We then awaited Apple's final review and approval.
Much to our surprise we received an email from Apple claiming our app had another 3.0 problem. Upon reading the report, we discovered that the Apple tester had not reported a crash or other showstopper, but instead had made a determination that a specific feature did not work the way he/she had thought it should work. And since they attributed this to iPhone OS 3.0, they rejected our update once again. The feature that was misinterpreted had not changed since the last time Apple approved our app. In fact, we regressed their report under 2.x and 3.0 and the feature worked identically in both configurations.
So beware, if your app update ends up in the hands of an over zealous Apple tester who, in my opinion, oversteps his/her responsibilities, you may find yourself with more delays in getting your app approved.