fortran: Fix PMPI interface bugs in mpi_f08 module

[kawashima-fj]

This PR brings bug fixes which were cared in MPI_ interface but were not cared in PMPI_ interfaces of the Fortran mpi_f08 module.

As the result, ompi/mpi/fortran/use-mpi-f08/mod/pmpi-f08-interfaces.F90 with sed -e s/PMPI_/MPI_/g -e s/pmpi_/mpi_/g becomes same as ompi/mpi/fortran/use-mpi-f08/mod/mpi-f08-interfaces.F90 except copyright statements.

@jsquyres, you changed PMPI_Aint_add_f08 and PMPI_Aint_diff_f08 to subroutines from functions in your 258d1aa. I supposed it was an unintentional change and I reverted them to functions in my d4fc404. Could you confirm it just in case?

Better fix is generating pmpi-* files from mpi-* files. I remain it as a future work.

I'll create PRs for release branches to handle this PR and #4659.

NEWS: Correct PMPI_Aint_{add|diff} to be functions (not subroutines) in the Fortran mpi_f08 module.

[bwbarrett]

What are the ABI impacts of this change (if any, I suck at Fortran)?

[jsquyres]

@bwbarrett It breaks ABI for PMPI_Aint_{add|diff} (by changing them from subroutines to functions). But the OMPI was wrong (i.e., the MPI spec says they must be functions). Therefore, no user applications could have been using these functions.

This is worth a note in NEWS, though.

[kawashima-fj]

@bwbarrett @jsquyres My commit message d4fc404 explains it.

Strictly speaking, this change involves binary incompatibility.
But nobody used these subroutines and nobody will be affected because
these subroutines were useless (didn't return a calculated value).

Is it OK to merge this change to release branches (v2.0, v2.1, v3.0, v3.1)?

The NEWS item will be something like this:

• Correct PMPI_Aint_{add|diff} to be functions (not subroutines) in the Fortran mpi_f08 module.

[bwbarrett]

@kawashima-fj, yes, I think we should merge to the impacted release branches (although v2.0.x is dead, so don't worry about that branch).