Resolve "Reproductibility issue with nn_comm=1 + flux form momentum advection schemes using small mpp domain sizes"

Set default surface roughness option for gls and tke to nn_mxlice=0
(instead of 2), i.e. "no ice" case. Switched on the option in reference
configurations that do have sea-ice.

Ice thickness dependent roughness in GLS and TKE

See merge request nemo/nemo!49

(cherry picked from commit 7c412688)

79aab267 Ice thickness dependent roughness in GLS and TKE

This reverts commit b6f56da1, reversing
changes made to 32318db1.

fix run.stat writing logical for monoproc case
force prt_ctl + run.stat flushing when using monoproc

Remove 0.5 factor on zmod, already accounted for in ztx zty

Remove silly change on mbku/mbkv/mbkf at the surface: this is useless
and is not compatible with 1 layer configurations (e.g. SWG and
ICE_AGRIF). Revert changes to namelists in ICE_AGRIF

Fix istate.F90 to avoid access to unallocated arrays in the SWE case. This change enables the SWG configuration to pass SETTE tests

Guillaume Samson authored 3 years ago and Sebastien Masson committed 3 years ago

new expression to compute air and surface potential temperatures in bulk formulae (old ticket #2632)