Split marbl_mod and marbl_ciso_mod

[mnlevy1981]

Separate the initialization, surface flux computation, and interior tendency computations in both marbl_mod.F90 and marbl_ciso_mod.F90: the end result is replacing those two modules with seven new ones and moving some routines to marbl_init_mod.F90:

1. marbl_surface_flux_mod -- contains marbl_surface_flux_compute(), formerly marbl_set_surface_forcing()
2. marbl_interior_tendency_mod -- contains marbl_interior_tendency_compute(), formerly marbl_set_interior_forcing(). Also has marbl_set_global_scalars_interior(), which needs to be renamed (ideally it would be moved to marbl_glo_avg_mod but it calls adjust_bury_coeff() which needs to stay in marbl_interior_mod)
3. marbl_glo_avg_mod -- contains marbl_glo_avg_count_vars() and marbl_glo_avg_init_rmean_vals()
4. marbl_ciso_init_mod -- contains marbl_ciso_init_tracer_metadata()
5. marbl_ciso_surface_flux_mod -- contains marbl_ciso_surface_flux_compute(), formerly marbl_ciso_set_surface_forcing()
6. marbl_ciso_interior_tendency_mod -- contains marbl_ciso_interior_tendency_compute(), formerly marbl_ciso_set_interior_forcing().
7. marbl_interior_share_mod -- contains marbl_interior_share_update_sinking_particle_from_level_above(), which used to be a private routine (marbl_update_sinking_particle_from_prior_level()) in both marbl_mod and marbl_ciso_mod; now both tracer modules can use the same code. Perhaps this would also be a good place for adjust_bury_coeff(), allowing marbl_set_global_scalars_interior() to be moved to marbl_glo_avg_mod?

For consistency, the interface routines set_surface_forcing() and set_interior_forcing() have been renamed surface_flux_compute() and interior_tendency_compute(), respectively, and renames marbl_comp_nhx_surface_emis() to marbl_nhx_surface_emis_compute().

Addresses #86

[mnlevy1981]

Still to do: clean up the module-level comments in all these new modules. The original comments from marbl_mod.F90 ended up in marbl_surface_flux_mod.F90, and the comments from marbl_ciso_mod.F90 are in marbl_ciso_surface_flux_mod.F90 . It's not entirely clear to me where documentation like the following belongs:

  !  Multispecies ecosystem based on Doney et al. 1996, Moore et al., 2002
  !  Based on POP Global NCAR Nitrogen Ecosystem Model
  !  version 0.0 (June 15th, 1998) from S.C. Doney.
  !  Based on Doney et al., 1996 model.
  !  Climate and Global Dynamics, NCAR
  !  (doney@whoi.edu)
  !
  !  Version 1.0
  !  Multispecies, multiple limiting nutrient version of ecosystem
  !  based on mixed layer model of Moore et al.(2002).  Implemented here with
  !  fixed elemental ratios and including only the diatoms and small
  !  phytoplankton, with a parameterization of calcification,
  !  by Keith Lindsay and Keith Moore, Fall 2001 - Spring 2002.
  !  Calcification parameterization based on Moore et al. 2002.
  !
  !  Version 2.0, January 2003
  !    Adds diazotrophs as a phytoplankton group, (based on Moore et al., 2002a)
  !    Allows for variable fe/C for all phytoplankton groups
  !     Allows for variable si/C for the diatoms
  !     Adds explicit tracers for DON, DOP, DOFe
  !     variable remin length scale for detrital soft POM and bSi f(temperature)
  !     Extensive modifications to iron scavenging parameterization
  !     Addition of a sedimentary dissolved iron source,
  !        (implemented in ballast code as excess remin in bottom cell)
  !        coded by J.K. Moore, (jkmoore@uci.edu)
  !
  !   Version 2.01. March 2003
  !     corrected O2 bug
  !     corrected grazing parameter z_grz bug at depth
  !     dust dissolution at depth releases iron,
  !     increased length scale for dust diss., increased hard fraction dust
  !     no deep ocean reduction in scavenging rates,
  !     increase bSi OC/ballast ratio 0.3 -> 0.35,
  !     corrected bug in diazotroph photoadaptation, and diat and sp adapatation
  !
  !   Version 2.02.
  !     corrected bug in Fe_scavenge (units for dust), May 2003
  !     changed C/N/P ratios to 117/16/1 (Anderson & Sarmiento, 1994)
  !
  !   Version 2.03., July 2003
  !     Remin of DOM no longer temperature dependent,
  !     new iron scavenging parameterization added,
  !     some dissolution of hard fraction of ballast materials added
  !
  !   Version 2.1, September 2003
  !     modfied iron scavenging and dust dissolution at depth
  !
  !   Version 2.11, March 2004
  !     fixed bug in iron scavenging code, replace dust and POC flux_in w/ flux_out
  !
  !   Version 2.12, April 2004 - Final version for GBC paper revision,
  !     (Questions/comments, Keith Moore - jkmoore@uci.edu
  !
  !   References
  !   Doney, S.C., Glover, D.M., Najjar, R.G., 1996. A new coupled, one-dimensional
  !   biological-physical model for the upper ocean: applications to the JGOFS
  !   Bermuda Time-Series Study (BATS) site. Deep-Sea Res. II, 43: 591-624.
  !
  !   Moore, JK, Doney, SC, Kleypas, JA, Glover, DM, Fung, IY, 2002. An intermediate
  !   complexity marine ecosystem model for the global domain. Deep-Sea Res. II, 49:
  !   403-462.
  !
  !   Moore, JK, Doney, SC, Glover, DM, Fung, IY, 2002. Iron cycling and nutrient
  !   limitation patterns in surface waters of the world ocean. Deep-Sea Res. II,
  !   49: 463-507.

Also, there are two things I want to look into doing:

1. Split marbl_diagnostics_mod.F90 into marbl_diagnostics_mod.F90 and marbl_ciso_diagnostics_mod.F90
2. Always call marbl_ciso_*() routines, but have if (.not. ciso_on) return at the top... i.e. move the if (ciso_on) checks into the Ciso routines. I think that will be a cleaner template to follow when adding additional tracer modules such as Niso.

[mnlevy1981]

I've split marbl_diagnostic_mod.F90, and now realize that there are lots of variables, data types, and subroutines still using surface_forcing and interior_forcing -- I will be renaming all of those to surface_flux and interior_tendency, respectively.

[mnlevy1981]

To clarify the last comment, I want to replace surface_forcing with surface_flux in relationship to the computation done in marbl_surface_flux_compute() (formerly set_surface_forcing()). Datatypes and subroutines dealing with forcing fields requested by MARBL (u10_sqr, sst, Salinity, etc) should keep forcing in its moniker, and we do separate those forcings based on whether they apply to the surface flux computation or the interior tendency computation. Maybe those should be referred to forcings_for_surface_flux and forcings_for_interior_tendency? That's starting to get pretty long, especially for things like marbl_forcings_for_surface_flux_indexing_type and marbl_forcings_for_interior_tendency_indexing_type

[mnlevy1981]

Having the if (ciso_on) check in marbl_ciso_surface_flux_mod means I'm always trying to pass surface_input_forcings(ind%d1{3,4}c_id), and those indices are 0 for non-ciso runs. So I think we should pass all of surface_input_forcings (as well as ind) and let the subroutine pull out sst, d13c, and d14c.