a few missing functions for fv3jedi variational analysis in rrfs

[TingLei-NOAA]

This issue is to list and track a few missing pieces required for a fv3jedi variational process completely functioning as expected.
Some of them have been dealt with by PRs while still pending.

1. filling missing values in the interpolation of the state variable in the outer blocks when those filtering (or B) points are outside of the model domain.
   https://github.com/JCSDA-internal/saber/pull/1189
2. update 3d pressure after each outer loop
   https://github.com/JCSDA-internal/fv3-jedi/pull/1381 ( by @rtodling)
3. Usage of 3d pressure as an analysis variable while surface pressure is the control variable in both static B and ensemble B.
   The former seems had been or being dealt with by works inside gsibec. For ensemble B, code changes are needed.