Changes to OpenACC directives

Converting acc kernels to an acc parallel region, following earlier discussions
to only use kernels directive for temporary code regions.

Using the default(present) clauses for parallel regions to be consistent with
the rest of the code.

Author: abishekg7

src/core_atmosphere/dynamics/mpas_atm_boundaries.F

@@ -215,12 +215,14 @@ subroutine mpas_atm_update_bdy_tend(clock, streamManager, block, firstCall, ierr
         index_qv = index_qv_ptr
 
         ! Compute lbc_rho_zz
+        !$acc parallel default(present)
+        !$acc loop vector
+        do k=1,nVertLevels
+            zz(k,nCells+1) = 1.0_RKIND          ! Avoid potential division by zero in the following line
+        end do
+        !$acc end parallel
 
-        !$acc kernels
-        zz(:,nCells+1) = 1.0_RKIND          ! Avoid potential division by zero in the following line
-        !$acc end kernels
-
-        !$acc parallel
+        !$acc parallel default(present)
         !$acc loop gang vector collapse(2)
         do iCell=1,nCells+1
             do k=1,nVertLevels
@@ -230,7 +232,7 @@ subroutine mpas_atm_update_bdy_tend(clock, streamManager, block, firstCall, ierr
         !$acc end parallel
 
         ! Average lbc_rho_zz to edges
-        !$acc parallel
+        !$acc parallel default(present)
         !$acc loop gang worker
         do iEdge=1,nEdges
             cell1 = cellsOnEdge(1,iEdge)
@@ -244,7 +246,7 @@ subroutine mpas_atm_update_bdy_tend(clock, streamManager, block, firstCall, ierr
         end do
         !$acc end parallel
 
-        !$acc parallel
+        !$acc parallel default(present)
         !$acc loop gang vector collapse(2)
         do iEdge=1,nEdges+1
             do k=1,nVertLevels