* They are generated by cfortran.h with extern scope.

This lead to some NANQ when r was not initialized to 0 in connection
with the ICC(p) preconditioner.

* This will only be needed until Fortran 2003 C bindings are adopted.

* Error handling functions and such should now use PPM_default_comm instead
  of MPI_COMM_WORLD.

Clearing the boundary of r inside the icc(p) preconditioner is more secure
but had an negative impact on performance. Therefore the solver now makes
sure that the boundary of r is initialized to 0 before calling a preconditioner
on it.

* Also fixes a bug when building without ParMeTiS support.

* Also rename corresponding C sources.

* This is only one step mandated by the brain-dead behaviour of the
  NAG and PGI compilers respectively.

- Removed unnecessary, misleading comments
- initialized boundary of s