Fix for specified ice layout non repro issue

- Closes issue NOAA-GFDL#199
- Specified-Ice mode needs albedo update at every timestep because coszen changes every timestep
- Fix ice_layout non-repro of amip using specified ice mode
- Ensure Ice%sCS is associated before checking Ice%sCS%specified_ice

src/ice_model.F90

@@ -1269,12 +1269,16 @@ subroutine set_fast_ocean_sfc_properties( Atmos_boundary, Ice, IST, Rad, FIA, &
     Ice%t_surf(i2,j2,k2) = US%C_to_degC*Rad%t_skin(i,j,k) + T_0degC
   enddo ; enddo ; enddo
 
+  ! coszen has changed if we are using the specified ice mode
+  if (associated(Ice%sCS)) then
+    if(Ice%sCS%specified_ice) coszen_changed = .true.
+  endif
   ! set_ocean_albedo only needs to be called if do_sun_angle_for_alb is true or
   ! if the coupled model's radiation timestep is shorter than the slow coupling
   ! timestep.  However, it is safe (if wasteful) to call it more frequently.
   if (Rad%do_sun_angle_for_alb) then
     call set_ocean_albedo_from_astronomy(Ice, G, Time_start, Time_end)
-  elseif (coszen_changed .or. Ice%sCS%specified_ice) then
+  elseif (coszen_changed) then
     call set_ocean_albedo_from_coszen(Ice, G, Rad%coszen_nextrad)
   endif