Remove zorl interstitial variables

[climbfuji]

This PR removes the zorl (surface roughness length) interstitial variables and uses the persistent variables instead. This reduces memory footprint and computational overhead.

In order to maintain b4b reproducibility with all existing regression tests, commit edc9194. Without this change, the two coupled model runs (both fractional and binary mask) using a high-resolution ocean model have slightly different results (all other tests still pass). The b4b mismatch is due to a number of isolated points over open ocean (i.e. far away from the coast) with ice fractions close to one (but not exactly one, around 0.995). I believe that the requirement to reset those points to huge values for zorl is incorrect, that this commit edc9194 should be reverted, and that the baselines for the two coupled, high-res ocean runs should be updated. @DeniseWorthen helped me investigate these differences (thanks!).

Fixes #552

Associated PRs:

#596
NOAA-EMC/fv3atm#261
ufs-community/ufs-weather-model#479

For regression testing, see ufs-community/ufs-weather-model#479.

[ShanSunNOAA]

I approve it. Thanks for cleaning up, -Shan

[llpcarson]

looks good to me, approved

[grantfirl]

Looks fine to me. Why the file permissions change on module_sf_exchcoef.f90?

[climbfuji]

Looks fine to me. Why the file permissions change on module_sf_exchcoef.f90?

This file had the executable permissions set (755), but a Fortran source file isn't an executable, therefore rw-r-r (=644) is correct.

[grantfirl]

Looks fine to me. Why the file permissions change on module_sf_exchcoef.f90?

This file had the executable permissions set (755), but a Fortran source file isn't an executable, therefore rw-r-r (=644) is correct.

Makes perfect sense. Thanks. I guess the file permissions were set to 755 in error at some time in the past then and this file was the only one in ccpp-physics?

[climbfuji]

Looks fine to me. Why the file permissions change on module_sf_exchcoef.f90?

This file had the executable permissions set (755), but a Fortran source file isn't an executable, therefore rw-r-r (=644) is correct.

Makes perfect sense. Thanks. I guess the file permissions were set to 755 in error at some time in the past then and this file was the only one in ccpp-physics?

We fixed several of those over time, but they keep coming back. Assuming that some users have their editors set up wrong and that the files get the executable flags automatically. We need to watch out for this in future PRs.