[Mitgcm-support] changes needed for SGI MPI

[mitgcm-support at dev.mitgcm.org]

Modifications needed in ini_procs.F for code to work 
under SGI MPI:

 85% diff ini_procs.F.original ini_procs.F
297c297
<        CALL MPI_TYPE_VECTOR(elCount,elLen,elStride,MPI_REAL4,
---
>        CALL MPI_TYPE_VECTOR(elCount,elLen,elStride,MPI_REAL,
316c316
<        CALL MPI_TYPE_VECTOR(elCount,elLen,elStride,MPI_REAL8,
---
>        CALL MPI_TYPE_VECTOR(elCount,elLen,elStride,MPI_DOUBLE_PRECISION,
385c385
<        CALL MPI_TYPE_CONTIGUOUS(elCount,MPI_REAL4,
---
>        CALL MPI_TYPE_CONTIGUOUS(elCount,MPI_REAL,
403c403
<        CALL MPI_TYPE_CONTIGUOUS(elCount,MPI_REAL8,
---
>        CALL MPI_TYPE_CONTIGUOUS(elCount,MPI_DOUBLE_PRECISION,