A lot of times I will be writing a unit test which repeatedly calls some API that I have marked @CheckForNull. In the test I am not going to bother checking for null after each call; the method should not in fact return null in this case, and if it does, the test will catch that in some other way, such as an assertion or a NullPointerException.

Or test code may legitimately do all sorts of things that would be considered inappropriate in production code, like repeatedly sleeping for 100ms, or hardcoding a path name, etc.

Since it is typically of little value to run FindBugs on test code, and it introduces a lot of spurious warning badges, I would recommend disabling the automatic scan on test source roots (or at least doing so optionally).