+Note [Inline candidates]
+~~~~~~~~~~~~~~~~~~~~~~~~
+At one point I gave is_con_app a higher score than inline-candidate,
+on the grounds that "it's *really* helpful if dictionaries get inlined fast".
+However a nofib run revealed no change if they were swapped so that
+inline-candidate has the higher score. And it's important that it does,
+else you can get a bad worker-wrapper split thus:
+ rec {
+ $wfoo x = ....foo x....
+
+ {-loop brk-} foo x = ...$wfoo x...
+ }
+But we *want* the wrapper to be inlined! If it isn't, the interface
+file sees the unfolding for $wfoo, and sees that foo is strict (and
+hence it gets an auto-generated wrapper. Result: an infinite inlining
+in the importing scope. So be a bit careful if you change this. A
+good example is Tree.repTree in nofib/spectral/minimax. If is_con_app
+has the higher score, then compiling Game.hs goes into an infinite loop.
+