Disable failing RSpec/ExpectInHook cop for now

This can always be refactored later in a separate PR if anyone wants.

On IRC, the reason given by ewoud for this line was...

my idea was to verify that it was actually stubbing the fact - for
example, if it moves to os.family then you'd need to modify the stub
but if you're on Red Hat (like Fedora) then as a developer you don't
notice it