Missing CICE halo update when `calc_strair=.false.`

[kieranricardo]

It looks like CICE is missing a halo update when calc_strair=.false.: https://github.com/ACCESS-NRI/CICE/blob/1a6ffaaf34de370ed1d58205feb0fb7a5f63ff43/cicecore/cicedyn/dynamics/ice_dyn_eap.F90#L266

This affects the interpolation of the windstress from the T grid to the U grid. And can seen by comparing windstresses from the top and second rows in the CICE grid:

[kieranricardo]

Fixed with ACCESS-NRI/CICE@d21b801 ACCESS-NRI/CICE#48

[anton-seaice]

Does the issue also occur at the edge of other blocks ? Presumably every computational block has missing data then at the top and right hand edges if the strax/stray is missing from the halo cells

I think that coupling with NEMO uses a grid where the tripole closes on tracer points (rather than vector points), which means that every vector has tracer points around it.

[anton-seaice]

A 2d plot of strax and stray presumably show lines through them at the edges of the computational blocks ?