[codex] Allow ocean-only sea ice model without atmosphere

[taimoorsohail]

Summary

• allow OceanSeaIceModel no-atmosphere runs to use typed zero rainfall and snowfall inputs
• route ocean and sea-ice net flux assembly through dispatch instead of dereferencing atmosphere state directly
• preserve the existing NoAtmosInterfaceModel no-op interface flux behavior for ocean-only sea-ice coupling

Why

The constructor path already allows atmosphere=nothing, but the net-flux assembly still assumed an atmosphere exchanger and crashed when prognostic sea ice was present. This change makes the no-atmosphere path consistent with the existing type-based coupling design.

Closes #268

Validation

• Local tests not run, per repository instructions to leave test execution to GitHub CI.

[codecov]

Codecov Report

❌ Patch coverage is 77.77778% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/EarthSystemModels/components.jl	0.00%	2 Missing ⚠️
src/Oceans/assemble_net_ocean_fluxes.jl	81.81%	2 Missing ⚠️
...s/InterfaceComputations/atmosphere_ocean_fluxes.jl	88.88%	1 Missing ⚠️
...InterfaceComputations/atmosphere_sea_ice_fluxes.jl	88.88%	1 Missing ⚠️
src/EarthSystemModels/earth_system_model.jl	50.00%	1 Missing ⚠️
src/SeaIces/assemble_net_sea_ice_fluxes.jl	66.66%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[taimoorsohail]

OK the hard-coded assumptions of valid atmosphere stretch a lot further than I anticipated. This will take a bit to work on and merge, especially wrangling the weird CUDA errors.

[simone-silvestri]

I think we only need to change the assemble functions right? The flux computation should return nothing if there is no atmosphere.

[navidcy]

Is this captured by any test or doctest?

[navidcy]

Is this captured by any test or doctest?

a92ff45 added one