Fixing issue with SSMIS models

[IFenton]

Closes #331

[github-actions]

Coverage report

This PR does not seem to contain any modification to coverable code.

[jemrobinson]

LGTM. Is this the only issue with them though? Have you tested?

[IFenton]

LGTM. Is this the only issue with them though? Have you tested?

I haven't tested all of them, but it now runs for the quick_test model, and I'm just checking the naive_unet_naive model

The one I was least sure of was the reproject_unet_naive model. Is this the correct fix for that one?

[jemrobinson]

The one I was least sure of was the reproject_unet_naive model. Is this the correct fix for that one?

Yes, I think so, but we're unlikely to use that in any case (it's very slow and we're reprojecting at creation-time now).

[IFenton]

For reference, here is a run of the quick_test model for the SSMIS data: https://wandb.ai/turing-seaice/evaluate/runs/7782f286?nw=nwuserifenton

[IFenton]

Also runs fine for a CNN_UNet_CNN model: https://wandb.ai/turing-seaice/evaluate/runs/u2ea98af, so going to merge this now