Science/Cleanup: Verify handling of surface stresses for PBL scheme vs. diffusion solver

[jimmielin]

Split off from HB/diffusion solver review at #270 (comment)

In CAM's vertical_diffusion.F90, there is a note that

  case ( 'HB', 'HBR' )

     ! Modification : We may need to use 'taux' instead of 'tautotx' here, for
     !                consistency with the previous HB scheme.

the surface stresses passed to HB, the PBL scheme, are tautotx and tautoty, defined as tautot = cam_in%ws + tautms + taublj;

however, this is inconsistently handled compared to the diffusion solver (previously compute_vdiff), which use taux and tauy, defined as taux = cam_in%wsx (HB, UW) or taux = 0 (CLUBB_SGS)

maybe the handling here could be refactored in the future. For now, to separate these inconsistent definitions of surface stresses, the suffixes _for_holtslag_boville_boundary_layer_scheme and _for_vertical_diffusion have been introduced to make this artificial distinction.

[jimmielin]

This will be resolved mostly in #370