Explore the delz update in fv3 RRFS

[TingLei-NOAA]

This issue is intended to facilitate discussion on exploring the analysis of dz in regional fv3 data assimilation.
DZ (as well as delp) is prognostic variables in fv3 and used in the warm-start initialization (while no ps). We already have a clear and agreed-upon approach approach to be applied for regional fv3jedi to handle delp versus ps as discussed in #518.

Previously, Wanshu Wu proposed an a posteriori ("off-line") update procedure for dz. In principle, this approach:

1. calculate differences in geometric heights before and after analysis using analysis variables of ps and other variables ont directly related to pressure, and
2. calcuate the corresponding differences in the diagnostic dz which could be derived from the geometrical heights and use this differences as the increment of dz induced from the data assimilation.

@xyzemc and Ting had implemented this scheme (to be turned on by the option l_reg_update_hydro_delz) and Xiaoyan 's initial evaluation demonstrated promising results.
However, this option was not turned on in the current rrfsv1. Therefore, the following questions arise:

1. should we consider further evaluate this option, with the goa to use it in the future GSI based rrfs data assimilation ?
2. Should we also implement this capability in fv3jedi?

[ShunLiu-NOAA]

@TingLei-NOAA Thank you for reminding me about the delz option. Since it is not used in RRFSv1, testing it in FV3-JEDI is not currently a top priority. However, we should revisit this once the FV3-JEDI-based RDAS is established.