Create scheme for cam_thermo_water_update using dependency method

[jimmielin]

cam_thermo_water_update has to run a second time (the first time is in dp_coupling) before dycore_energy_consistency_adjust scheme runs, to provide updated generalized cp:

For both CAM4/5/6 and CAM7:

! in tphysac:

    !
    ! update cp/cv for energy computation based in updated water variables
    !
    call cam_thermo_water_update(state%q(:ncol,:,:), lchnk, ncol, vc_dycore,&
         to_dry_factor=state%pdel(:ncol,:)/state%pdeldry(:ncol,:))

! ... physics_dme_adjust

    ! ccpp-ized scheme dycore_energy_consistency_adjust

cam_thermo_water_update is currently in src/data/cam_thermo.F90. It calls
water_composition_update in src/data/air_composition.F90, which depends on

• get_cp_1hd (which depends on get_cp_dry_1hd) in air_composition;
• get_R_1hd (which depends on get_R_dry_1hd) in air_composition;
• a variety of species number and index information in air_composition, i.e., thermodynamic_active_species_num, thermodynamic_active_species_idx, dry_air_species_num, thermodynamic_active_species_cp, thermodynamic_active_species_R

Because air_composition properties and subroutines are also dependent on the dycore, it may not be feasible to CCPPize the entire cam_thermo_water_update and its related subroutines.

A dependent scheme can be made that depends on src/data/cam_thermo.F90 and calls the underlying cam_thermo_water_update with the to_dry_factor (note: the call from the dycore does not include this factor.) which would allow the CAM4/5/6/7 suites to run properly until the CAM thermo infrastructure can be completely CCPPized (if in the future)