Fix CI review findings, build errors, and remaining issues

- Fix bubs_glb=0 placement after declarations (Fortran standard violation)
- Add missing max_dt declaration and GPU_DECLARE in m_time_steppers.fpp
- Add @:PROHIBIT for particle restart without MPI
- Deallocate beta_vars in pre_process and post_process finalizers
- Change vL_field/vR_field intent(inout) to intent(in) in s_gradient_field
- Fix bare integer 3 -> 3._wp in m_bubbles.fpp
- Fix NVTX typo RMDA -> RDMA

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: sbryngelson

src/common/m_mpi_common.fpp

@@ -388,12 +388,12 @@ contains
         real(wp), intent(out) :: Rc_min_glb
         integer, intent(out) :: bubs_glb
 
-        bubs_glb = 0
-
 #ifdef MFC_SIMULATION
 #ifdef MFC_MPI
         integer :: ierr !< Generic flag used to identify and report MPI errors
 
+        bubs_glb = 0
+
         ! Reducing local extrema of ICFL, VCFL, CCFL and Rc numbers to their
         ! global extrema and bookkeeping the results on the rank 0 processor
         call MPI_REDUCE(icfl_max_loc, icfl_max_glb, 1, &
@@ -417,6 +417,7 @@ contains
 #else
 
         icfl_max_glb = icfl_max_loc
+        bubs_glb = 0
 
         if (viscous) then
             vcfl_max_glb = vcfl_max_loc
@@ -1341,7 +1342,7 @@ contains
                     call nvtxStartRange("BETA-COMM-DEV2HOST")
                     $:GPU_UPDATE(host='[buff_send]')
                     call nvtxEndRange
-                    call nvtxStartRange("BETA-COMM-SENDRECV-NO-RMDA")
+                    call nvtxStartRange("BETA-COMM-SENDRECV-NO-RDMA")
 
                     call MPI_SENDRECV( &
                         buff_send, buffer_count, mpi_p, dst_proc, send_tag, &

src/post_process/m_global_parameters.fpp

@@ -1108,6 +1108,7 @@ contains
 #endif
 
         if (allocated(neighbor_ranks)) deallocate (neighbor_ranks)
+        if (allocated(beta_vars)) deallocate (beta_vars)
 
     end subroutine s_finalize_global_parameters_module
 

src/pre_process/m_global_parameters.fpp

@@ -1114,6 +1114,7 @@ contains
 #endif
 
         if (allocated(neighbor_ranks)) deallocate (neighbor_ranks)
+        if (allocated(beta_vars)) deallocate (beta_vars)
 
     end subroutine s_finalize_global_parameters_module
 

src/simulation/m_bubbles.fpp

@@ -597,7 +597,7 @@ contains
                             do l = 1, num_dims
                                 f_bTemp = f_get_bubble_force(fPos(l), fR, fV, fVel(l), fmass_g, fmass_v, &
                                                              fRe, fRho, cell, l, q_prim_vf)
-                                aTemp(l) = 2._wp*f_bTemp/(fmass_g + fmass_v) - 3*fV*fVel(l)/fR
+                                aTemp(l) = 2._wp*f_bTemp/(fmass_g + fmass_v) - 3._wp*fV*fVel(l)/fR
                             end do
                             do l = 1, num_dims
                                 fVel(l) = fVel(l) + h*aTemp(l)

src/simulation/m_particles_EL.fpp

@@ -603,7 +603,9 @@ contains
         real(wp) :: file_time, file_dt
         integer :: file_num_procs, file_tot_part, tot_part
 
-#ifdef MFC_MPI
+#ifndef MFC_MPI
+        @:PROHIBIT(.true., "Lagrangian particle restart requires MPI (--mpi)")
+#else
         real(wp), dimension(20) :: inputvals
         integer, dimension(MPI_STATUS_SIZE) :: status
         integer(kind=MPI_OFFSET_KIND) :: disp
@@ -778,10 +780,10 @@ contains
         real(wp) :: myMass, myR, myBeta_c, myBeta_t, myR0, myRe, mydrhodt, myVolumeFrac, myGamma, rmass_add, func_sum
         real(wp), dimension(3) :: myVel, myPos, force_vec, s_cell
         logical :: only_beta
-        only_beta = .false.
-
         integer :: k, l, i, j
 
+        only_beta = .false.
+
         if (lag_params%pressure_force .or. lag_params%added_mass_model > 0) then
             do l = 1, num_dims
                 if (l == 1) then
@@ -1650,10 +1652,11 @@ contains
         integer :: patch_id, newBubs, new_idx
         integer, dimension(3) :: cell
         logical :: inc_ghost
-        inc_ghost = .false.
         real(wp) :: myR, func_sum
         real(wp), dimension(3) :: myPos, myVel, myForce
         logical :: only_beta
+
+        inc_ghost = .false.
         only_beta = .true.
 
         call nvtxStartRange("LAG-BC")
@@ -2030,8 +2033,8 @@ contains
     subroutine s_gradient_field(vL_field, vR_field, dq, dir, field_var)
 
         real(stp), dimension(idwbuff(1)%beg:, idwbuff(2)%beg:, idwbuff(3)%beg:), intent(out) :: dq
-        real(wp), dimension(idwbuff(1)%beg:, idwbuff(2)%beg:, idwbuff(3)%beg:, 1:), intent(inout) :: vL_field
-        real(wp), dimension(idwbuff(1)%beg:, idwbuff(2)%beg:, idwbuff(3)%beg:, 1:), intent(inout) :: vR_field
+        real(wp), dimension(idwbuff(1)%beg:, idwbuff(2)%beg:, idwbuff(3)%beg:, 1:), intent(in) :: vL_field
+        real(wp), dimension(idwbuff(1)%beg:, idwbuff(2)%beg:, idwbuff(3)%beg:, 1:), intent(in) :: vR_field
         integer, intent(in) :: dir, field_var
 
         integer :: i, j, k

src/simulation/m_time_steppers.fpp

@@ -70,14 +70,16 @@ module m_time_steppers
 
     real(wp), allocatable, dimension(:, :, :, :, :) :: rhs_mv
 
+    real(wp), allocatable, dimension(:, :, :) :: max_dt
+
     integer, private :: num_ts !<
     !! Number of time stages in the time-stepping scheme
 
     integer :: stor !< storage index
     real(wp), allocatable, dimension(:, :) :: rk_coef
     integer, private :: num_probe_ts
 
-    $:GPU_DECLARE(create='[q_cons_ts,q_prim_vf,q_T_sf,rhs_vf,q_prim_ts1,q_prim_ts2,rhs_mv,rhs_pb,rk_coef,stor,bc_type]')
+    $:GPU_DECLARE(create='[q_cons_ts,q_prim_vf,q_T_sf,rhs_vf,q_prim_ts1,q_prim_ts2,rhs_mv,rhs_pb,max_dt,rk_coef,stor,bc_type]')
 
 !> @cond
 #if defined(__NVCOMPILER_GPU_UNIFIED_MEM)