*Fix ALE_remap_scalar call with h_in_Z_units

  Corrected an argument to the call to ALE_remap_scalar() when h_in_Z_units is
true in MOM_initialize_tracer_from_Z(), to avoid the problems documented in
github.com//issues/589.  A comment was also added explaining the
logic of what is going on in this fork of the code.  This commit will change
answers with some generic tracers that are initialized from a Z-space input
file, restoring them to previous values that worked previously (before about
Feb. 1, 2024 on dev/gfdl) in Boussinesq configurations without dimensional
consistency testing, but in a new form that does pass the dimensional
consistency testing for depths and thicknesses.  All answers are bitwise
identical in any cases that do not use generic tracers.

src/initialization/MOM_tracer_initialization_from_Z.F90

@@ -200,8 +200,10 @@ subroutine MOM_initialize_tracer_from_Z(h, tr, G, GV, US, PF, src_file, src_var_
     enddo ; enddo
 
     if (h_is_in_Z_units) then
+      ! Because h is in units of [Z ~> m], dzSrc is already in the right units, but we need to
+      ! specify negligible thickness values with the right units.
       dz_neglect = set_dz_neglect(GV, US, remap_answer_date, dz_neglect_edge)
-      call ALE_remap_scalar(remapCS, G, GV, kd, hSrc, tr_z, h, tr, all_cells=.false., answer_date=remap_answer_date, &
+      call ALE_remap_scalar(remapCS, G, GV, kd, dzSrc, tr_z, h, tr, all_cells=.false., answer_date=remap_answer_date, &
                             H_neglect=dz_neglect, H_neglect_edge=dz_neglect_edge)
     else
       ! Equation of state data is not available, so a simpler rescaling will have to suffice,