diaar5: Suspicious use of INTEGER in the ocean Volume computation

Hello, The problem I mentioned in the title of this issue is related to the way vol0 and thick0(:,:) are calculated in dia_ar5_init in diaar5.F90 :

   SUBROUTINE dia_ar5_init
      !!----------------------------------------------------------------------
      !!                  ***  ROUTINE dia_ar5_init  ***
      !!
      !! ** Purpose :   initialization for AR5 diagnostic computation
      !!----------------------------------------------------------------------
      INTEGER  ::   inum
      INTEGER  ::   ik, idep
      INTEGER  ::   ji, jj, jk  ! dummy loop indices
      REAL(wp) ::   zztmp
      REAL(wp), ALLOCATABLE, DIMENSION(:,:,:,:) ::   zsaldta   ! Jan/Dec levitus salinity
      REAL(wp), ALLOCATABLE, DIMENSION(:,:)     ::   zvol0
..........
         area_tot  = glob_sum( 'diaar5', e1e2t(:,:) )

         ALLOCATE( zvol0(jpi,jpj) )
         zvol0 (:,:) = 0._wp
         thick0(:,:) = 0._wp
         DO_3D( 1, 1, 1, 1, 1, jpkm1 )   ! interpolation of salinity at the last ocean level (i.e. the partial step)
            idep = tmask(ji,jj,jk) * e3t_0(ji,jj,jk)
            zvol0 (ji,jj) = zvol0 (ji,jj) +  idep * e1e2t(ji,jj)
            thick0(ji,jj) = thick0(ji,jj) +  idep
         END_3D
         vol0 = glob_sum( 'diaar5', zvol0 )
         DEALLOCATE( zvol0 )
.....

Using idep (integer) truncate e3t_0 to an integer value.
I do not capture the rationale for that (if any).
The impact is just that vol0 and thick0(:,:) are not computed accurately...

This idep has been introduced in NEMO between 4.0.1 and 4.0.2