For regional fv3-jedi analysis with qoption2

[Masanori-NOAA]

This PR is for conducting regional fv3-jedi 3DVar analysis with qoption2. The specific code changes are as follows:

• (c2710e5) : Found that the external subroutine “genqsat” failed to handle array inputs properly, affecting the background error of qoption2 normalized RH which dependent on background RH. After refactoring genqsat into a module, the background saturated specific humidity and RH distributions now align with expectations (see figures below).

• (09e38be) : The oops generic interpolator does not extrapolate background values to analysis grid points located outside the LAM domain, leaving the values at these points undefined. To address this, added a process to fill these undefined points outside the model domain by copying background values from the nearest neighboring grid points.
• (15890a4) : Changes to enable the use of RRFSv1 berror_stats (with usenewgfsberror = .true.).
• (b5c88c6) : Fix related to 2nd outer loop crash in regional analysis
• (cab79d9) : Refactoring to remove redundant code

[rtodling]

This is to say that I am aware of this PR and will devote time to looking at it soon.

[rtodling]

Sorry: I am finally getting back to this and JEDI.

At first sight these changes look good.

I will run a quick test tomorrow and should be able to either approve - issue a tag - or provide any feedback needed.