grv function used by NST assumes wrong argument

[DeniseWorthen]

Description

The NST model uses the function grv to give the latitude dependence of gravity.

https://github.com/ufs-community/ccpp-physics/blob/c2ec9e5e165dac07a991424bd22bb0667cf32017/physics/module_nst_water_prop.f90#L485-L500

This function assumes the argument is the latitude but the value actually used by sfc_nst.f is the sin(latitude).

https://github.com/ufs-community/ccpp-physics/blob/c2ec9e5e165dac07a991424bd22bb0667cf32017/physics/sfc_nst.f#L372

Steps to Reproduce

Additional Context

See gravity forumla 1980 for the definition of gravity=f(latitude) used in this module.

[junwang-noaa]

Thanks to Denise for this bug fix. @yangfanglin @Qingfu-Liu @XuLi-NOAA FYI. Also would you like to do some testing to see the impact of this bug fix?

[yangfanglin]

maybe start with comparing results between two regression tests ? @XuLi-NOAA

[XuLi-NOAA]

maybe start with comparing results between two regression tests ? @XuLi-NOAA

I can do that once these two regression tests are available (or already, juts point them to me). As I understand, it purly/only affects the latittude dependent gravity accelaration variation, as Denise pointed, at 45 degree, the error due the bug is about 0.26%. And the impact of this parameter (g) on the NSST model must be very limited.

[yangfanglin]

@XuLi-NOAA making a few plots of NSST related variables and posting here should suffice for closing this PR