R21B to IC4 subroutine

[ErickRogers]

Short version:
Introduce R21B variant of Sice, as an option, to IC4 subroutine of WW3. This feature already exists in SWAN v41.31AB and in COAMPS WW3 v6.07.
Reference:
Rogers, W.E., Yu, J., Wang, D.W., 2021. Incorporating dependencies on ice thickness in empirical parameterizations of wave dissipation by sea ice, Technical Report, NRL/OT/7320-21-5145, 35 pp., https://www7320.nrlssc.navy.mil/pubs/index.php

Long version:
Along with R21B, I'm adding a model from Meylan et al. (2018)
The Meylan et al. model will be denoted IC4M8 and R21B will be denoted IC4M9 ("Methods" 8 and 9)
The Meylan model can also be activated as a sub-option of the "IC5" module. This redundancy is intentional.
Here are my inline notes:
! 8) Meylan et al. (JGR 2018), eq. 48. "Model with Order 3 Power
! Law". The is denoted as the "M2" model by Liu et al. (JPO 2020)
! It is a function of ice thickness and wave period.
! ki = ChfM2h_icefreq^3
! where ChfM2 is a coefficient of proportionality which formally
! includes viscosity, density, and gravity parameters, see
! Meylan et al. (JGR 2018) for details.
! ChfM2 has units of s3/m2
! It is equation 53 in Meylan et al. (2018) and equation 16 in
! Liu et al. (2020).
! This method is functionally the same as IC5M2 in WW3 and is
! redundantly included here as IC4M8 because it is in the same
! "family" as IC4M7 and IC4M9, being in the form of
! ki=Chf * h_ice^m * freq^n .
! Calibrations:
! * Liu et al. has ChfM2=eta*(2pi)^3/(10259.81^2)
! ** eta=14.0 for "Sikuliaq" case of Liu et al., so ChfM2=0.035
! ** eta=3.0 for "SIPEX" case of Liu et al., so ChfM2=0.0075
! * Rogers et al. (tech rep. 2021b, "R21B") :
! ** Fit to Rogers et al. (CRST 2021a "R21A") ChfM2=0.059 (SD)
! suggested default is marked with "(SD)", for consistency
! with SWAN v41.31AB
!
! 9) Rogers et al. (tech. rep. 2021b). The "monomial power fit"
! described in section 2.2.3. It is the general form above,
! ki=Chf * h_ice^m * freq^n
! but is constrained such that m=n/2-1.
! This is also given as equation 2 in Yu et al. (CRST 2022).
! * R21B, calibration to R21A: Chf=2.9 and n=4.5 (SD)
! * Yu et al. (2022) calibration to R21A : Chf=2.4 and n=4.46
! (noting that c_n=0.108 and Chf=c_n*(2*pi/sqrt(g))^n)
! * Yu (2022) adjusted the prior calibration to get better fit
! to higher frequency lab measurements and got:
! Chf=7.89 and n=4.8
! suggested default is marked with "(SD)", for consistency
! with SWAN v41.31AB