Stochastic physics cleanup

[pjpegion]

This pull request includes a massive cleanup of the stochastic physics code, addition of a new random number generator for the CA code which allows for the same answer for different PE layouts.
Adds restart capability for ca_global, and unit test programs in subdirectory unit_test, which currently only works on hera.
Also, iseed_ca is now an 8-bit integer, so the seed in the workflow can used the same methodology as iseed_sppt, etc.

This also fixes
#33
and
#25

[lisa-bengtsson]

The code addresses the CA MPI reproducibility issue:
ufs-community/ufs-weather-model#743
Provides a better solution for the CA restart as discussed:
ufs-community/ufs-weather-model#797
Fixes the CA GNU debug issue:
ufs-community/ufs-weather-model#770

[lisa-bengtsson]

Hi Phil! Thanks for this, a lot of work has gone in to this one clearly! Regarding the cleanup of the stochastic physics code, I don't know much about the capabilities of the removed code, so I don't have many comments on that piece. But I assume that you know if anyone outside of PSL would be dependent on any of those features?

Regarding the CA code, in update_ca.F90 with this line:
!Return if not allocated:
if(.not. allocated(board) .and. .not. allocated(lives) .and. .not. allocated(board_g) .and. .not. allocated(lives_g))return

What happens if you only run ca_sgs? Or only ca_global?

A small remark, in the global_ca part of update_ca.F90 , newcell=0 looks to be set twice.

It looks like there is a fv3atm submodule update in the stochastic_physics_wrapper.F90 (and atmos_model.F90?) that goes with this PR? Could be good to post that number in the description.

[pjpegion]

1. most of the deleted files have been moved into spectral_transforms.F90, and are still available.
2. re: code in update_ca.F90, I will take a look. It looked like things were fine if I ran only ca_global or ca_sgs, but will check again
3. I will remove the duplicate newcell = 0
4. I am going to merge in the ocean stochastic physics PR, which also touches fv3atm, so I haven't issued that PR yet. When I do, I will post it here.

[pjpegion]

the line
if(.not. allocated(board) .and. .not. allocated(lives) .and. .not. allocated(board_g) .and. .not. allocated(lives_g))return
will only return if nothing is allocated.

[pjpegion]

One thing to note is that all of the code clean-up with the spectral transforms does not change the baseline results for the stochastic physics or land perturbation tests.

[lisa-bengtsson]

I'm a little curious about the io_layout and need for the io domain specification, does it have to be separate from the ca compute domain?

[pjpegion]

That is needed for the new way of reading and writing restarts.

[pjpegion]

associated fv3atm PR is NOAA-EMC/fv3atm#396 and
ccpp-physics PR NCAR/ccpp-physics#737

[climbfuji]

Not a thorough review, I only skimmed through the code changes