Use VADER for insitu/potential TL/AD

[travissluka]

Description

• closes Use vader for temperature linear variable change #946

changes:

• the ctest test_soca_3dvar_godas now uses VertInterp instead of InsituTemperature and so should make use of VADER for the TL/AD
• added a bunch of hacky looking code to do some variable renaming to make VADER happy. Don't look at this too much yet, it will be cleaned up a bit before this PR is merged, and will go away entirely when a follow up PR is done to clean up the variable names.

@Dooruk you should be able to use this to test your vader branch. In order for it to compile, I added a changeVarAD() empty stub, and commented out all the code in changeVarTL() and changeVarTraj() because they currently contain bugs.

Dependencies

depends on matching VADER branch

[Dooruk]

Thanks @travissluka !

It's still not clear to me what atlas grid type would handle MOM6 tripolar grid (which structured type). For instance, ORCA has an add-on for atlas:

https://github.com/ecmwf/atlas-orca

Does tripolar MOM6 grid differ from ORCA dramatically that we don't need such add-on?

@danholdaway suggested using an atmospheric gaussian grid that VADER tests currently work for and just replace existing values with ocean variables. I can do that but i would like to have a permanent solution eventually.

[travissluka]

@Dooruk I agree with Dan that it's probably easier to just use an existing gaussian grid. I don't think we need to go through the trouble of making sure the same tri-polar grid works with the vader ctests