Several PFT types have confusing names and dimensions

[mnlevy1981]

1. Rename the following:

   • autotroph_secondary_species_type -> autotroph_derived_terms_type
   • zooplankton_secondary_species_type -> zooplankton_derived_terms_type
   • autotroph_type -> autotroph_settings_type
   • zooplankton_type -> zooplankton_settings_type

2. Move (autotroph_cnt, km) dimensions into autotroph_local_type and autotroph_derived_terms_type [same for (zooplankton_cnt, km) and the zooplankton equivalents]

   • Instead of declaring local variables inside interior_tendency_compute(), move variables into marbl_interface_class and construct them during init.
   • This should help with memory access as we implement migrate k loop inside of subroutines called from marbl_set_interior_forcing #53

[mnlevy1981]

To go along with the changes from autotroph_type -> autotroph_settings_type and zooplankton_type -> zooplankton_settings_type, I also changed grazing_type -> grazer_settings_type. This also led to variable name changes:

• autotrophs -> autotroph_settings
• zooplankton -> zooplankton_settings
• grazing -> grazer_settings

These changes propagate through the settings YAML files, MARBL python scripts, and POP wrappers to the MARBL python scripts (but don't change answers).

[mnlevy1981]

Rather than grazer_settings, I ended up renaming grazing to grazing_relationship_settings. This issue is addressed in marbl0.33.0