Add two halo updates for taux and tauy in mom_surface_forcing_nuopc

[breichl]

• In A and B grid configuration halos were never updated after taux/tauy were populated.
• This propagated through to the ustar_gustless field, hence caused a restart issue when using ustar_gustless in parameterizations.
• This appears to correct the restart issue (restart reproducibility (without waves) when USE_LA_LI2016=True #46 ) by updating the halos at the end of the A and B grid taux/tauy loops.

[jiandewang]

look good to me, will wait for confirmation from Denise on her RT runs

[DeniseWorthen]

@jiandewang I have tested on orion and I can report that:

1. this fix does not change existing non-benchmark baselines. So the fix has no impact unless we have USE_LA_LI2016=TRUE

2. this fix allows all non-wave benchmark tests to restart reproduce when used with USE_LA_LI2016=TRUE

[jiandewang]

@DeniseWorthen this means there is no need to create new baseline, right ?

[DeniseWorthen]

@jiandewang We want the benchmark RTs to track as closely as possible the benchmark configuration. The benchmarks use USE_LA_LI2016=TRUE but since waves are not restart repro, we set this variable false in the non-wave benchmark RTs. This allowed us to test and maintain restart repro for the non-wave benchmark config.

Now that we can set it true and still maintain restart repro, we want to modify the non-wave benchmark RTs to use it. This will change the baselines for all non-wave benchmark tests.

[jiandewang]

@DeniseWorthen got what you mean, so with this fixing we can set USE_LA_LI2016 to T for all RT cases. I am going to approval this PR. Thanks for you, Jessica and Brandon's quick work.

[jiandewang]

the fixing solved the restart repro issue.