- Added 2 new blockschedulers, a greedy algorithm and an "optimal" ILP that

- Added 2 new blockschedulers, a greedy algorithm and an "optimal" ILP that
  both try to transform as many jumps as possible to fallthroughs (weighted by
  execution frequency). The results are most of the times better than the extbb
  blockscheduler.
  Though it seems like x86 branch prediction sees conditional backward
  jumps as always taken, so we have to make sure that for 50/50 jumps we don't
  create backward jumps. (nothing implemented for this yet)