+Note [Interaction of exprIsCheap and lone variables]
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+The lone-variable test says "don't inline if a case expression
+scrutines a lone variable whose unfolding is cheap". It's very
+important that, under these circumstances, exprIsConApp_maybe
+can spot a constructor application. So, for example, we don't
+consider
+ let x = e in (x,x)
+to be cheap, and that's good because exprIsConApp_maybe doesn't
+think that expression is a constructor application.
+
+I used to test is_value rather than is_cheap, which was utterly
+wrong, because the above expression responds True to exprIsHNF.
+
+This kind of thing can occur if you have
+
+ {-# INLINE foo #-}
+ foo = let x = e in (x,x)
+
+which Roman did.
+