+Simplify rescaling by US%L_to_Z

[Hallberg-NOAA]

This PR consists of three commits that eliminate 46 instances where there was a need to multiply by US%L_to_Z during run-time. There are 9 instances where there are compensating additions of factors of US%L_to_Z to scale or conversion arguments that are used during model initialization.

The first commit introduces a newly rescaled version of the gravitational acceleration, GV%g_Earth_Z_T2, to the verticalGrid_type as a copy of GV%g_Earth that uses dimensional rescaling of [Z T-2 ~> m s-2] instead of [L2 Z-1 T-2 ~> m s-2]. GV%g_Earth_Z_T2 is more convenient for single-column energy calculations (for which horizontal scales should be irrelevant), but the two variables will only differ by an integer power of two when dimensional rescaling is applied. This new version of the gravitational acceleration is used in 15 files.

Another US%L_to_Z**2 rescaling factor was eliminated from the code in ePBL_column() by folding it into MKE_to_TKE_effic.

The second commit revises the internal rescaling of turbulent kinetic energies in MOM_bulk_mixed_layer to work in units of [H Z2 T-2] instead of [H L2 T-2], for greater consistency with the rescaling of TKE elsewhere. These changes included using GV%g_Earth_Z_T2 instead of GV%g_Earth in 16 places that contribute to potential energy calculations. In 5 lines, the unit scaling factors were eliminated, while in 2 others the rescaling factors were revised. In addition, the rescaling factors to go from the units of mean kinetic energy to those of turbulent kinetic energy were folded into the internal representation of the bulk Richardson numbers. The units in comments describing 58 variables and the conversion arguments for 10 diagnostics were updated accordingly.

The third commit rescales the internal representation of the FGNV_STRAT_FLOOR variable and thickness_diffuse%N2_floor in MOM_mixedlayer_restrat to include appropriate factors of US%Z_to_L to reflect the scaling of aspect ratios. This leads to the elimination of one rescaling factor in thickness_diffuse_full() and its replacement by another in a scale factor for a get_param() call.

All answers and output are bitwise identical, but there is a new element in a transparent type. The specific commits in this PR include:

• fd4d3baac Rescale strat_floor
• c9892e76d Revise rescaling in bulkmixedlayer
• bb91495f9 +Add alternate gravity variable GV%g_Earth_Z_T2