I think there is a bug in the `dissolve_floe` code as it indexes into the ocean as `[yidx, xidx]` and I think it should be the other way around. Additionally, the simulation became unstable. Not sure why. This will require an investigation.