Say that the ocean mesh comes first

[JanStreffing]

Draft, pending input from @fernandadialzira on the mesh generation pipeline.

This is the sixth item of #3, "how to modify an ocean restart". The answer for the deep time case turns out to be that you do not. Bathymetry and sea level are properties of the mesh, and the ocean is cold started from climatology on the new mesh rather than carried over from an edited restart.

The page previously said a coupled paleo run needs a FESOM2 mesh for the time slice and then never said where one comes from, which left the impression that ocp-tool produces it. It does not, and the ordering is the other way round: the mesh is built first and the OpenIFS land-sea mask is derived from it.

Deliberately short. The method is published open access in Matos et al. (2026), https://doi.org/10.1016/j.gloplacha.2025.105196, and ocp-tool PR JanStreffing/ocp-tool#47 is the Python implementation of its atmosphere side. Restating the anomaly method, the mesh constraints, the node counts or the spin up lengths here would be a second copy of a methods section that has just been properly written up, and the copy in these docs is the one nobody would update. So the section names what is in the paper and links it.

The open question, and the reason this is a draft: the last line says generating the mesh is not part of ocp-tool and that where that pipeline lives is not written down here. If Fernanda can say what she used and where it lives, that line becomes a pointer instead of an admission.

13 lines, between the 12 line environment section and the 19 line vegetation one. Build unchanged at 9 warnings, none from the page.