While I agree with the overall idea behind your post, your assumption that the activation lock can be defeated by sheer git and determination is silly. If implemented correctly (and I have no idea if it will be) it's very likely that it won't be bypassed.