It's very easy to have too many tests and lose track of what kind of feedback each one brings. Moreover, testing on multiple devices can get expensive, both in terms of CI cost and lost productivity waiting for tests to pass.