Alex Brown has noticed that there is a problem with some LFRic intergrid kernels that map continuous (W2) fields between fine and coarse meshes.

This seemed to happen when the model's halo depth is smaller than the number of fine cells per coarse cell. As the field is continuous, the model attempts to do a halo exchanges but we think they are not being updated to the correct depth.

For now our work-around (on a branch) is to change the meta-data to say that the mapped field is discontinuous. This prevents any halo exchanges calls from being inserted by PSyclone (as these kernels we don't actually need to do any halo exchanges!)

The offending kernels are:
https://code.metoffice.gov.uk/trac/lfric/browser/LFRic/trunk/gungho/source/kernel/prolong_w2_kernel_mod.F90
https://code.metoffice.gov.uk/trac/lfric/browser/LFRic/trunk/gungho/source/kernel/restrict_w2_kernel_mod.F90
https://code.metoffice.gov.uk/trac/lfric/browser/LFRic/trunk/gungho/source/kernel/weights_prolong_w2_kernel_mod.F90