Fix bug in multi-region SplitExplicitFreeSurface

[simone-silvestri]

I noticed there was this bug.
This meant that the correct fill halos was not hit for a cubed sphere, since it requires a tuple of fields with (Face, Center, <:Any) and (Center, Face, <:Any). defined here

Oceananigans.jl/src/MultiRegion/cubed_sphere_boundary_conditions.jl

Lines 345 to 347 in 5bbc2a1

	function fill_halo_regions!(field_1::CubedSphereField{<:Face, <:Center},
	field_2::CubedSphereField{<:Center, <:Face}; signed = true, kwargs...)
	grid = field_1.grid

@siddharthabishnu this might be the error that you were searching for? Can you test the baroclinic aquaplanet on this branch? If this fixes you issue can you also add a test?

[siddharthabishnu]

I noticed there was this bug. This meant that the correct fill halos was not hit for a cubed sphere, since it requires a tuple of fields with (Face, Center, <:Any) and (Center, Face, <:Any). defined here

Oceananigans.jl/src/MultiRegion/cubed_sphere_boundary_conditions.jl

Lines 345 to 347 in 5bbc2a1

	function fill_halo_regions!(field_1::CubedSphereField{<:Face, <:Center},
	field_2::CubedSphereField{<:Center, <:Face}; signed = true, kwargs...)
	grid = field_1.grid

@siddharthabishnu this might be the error that you were searching for? Can you test the baroclinic aquaplanet on this branch? If this fixes you issue can you also add a test?

It might very well be! This is a major bug! Great catch! I clearly remember implementing this correctly earlier, so I must have missed it later while resolving merge conflicts for another major PR. I’ll add a test.