Hand unrolled version of dslash in a separate class.

Useful to compare; raises Intel compiler from 9GFlop/s to 17.5 Gflops.
                   on ivybridge core. Raises Clang form 14.5 to 17.5

lib/Grid.h

@@ -65,7 +65,6 @@
 
 namespace Grid {
 
-
   void Grid_init(int *argc,char ***argv);
   void Grid_finalize(void);
   // internal, controled with --handle