Useful optimisation for set-cost-centre

If you compile, for example, drvrun014 with -prof -auto-all, you'll see stuff like

    (scc "c" (dataToTag#)) y

This generates bad code, because we end up eta-expaning dataToTag, which allocates an extra function closure.

We think that in general

   (scc "c" e) y  = scc "c" (e y)

to within a small constant factor. So maybe the simplifier, or CorePrep, or both, should do this transformation.