Outstanding technical changes to regridding code

[fionaRust]

This ticket records outstanding technical changes that we want to make to the regridding code following #1438. Some of these points may require discussion or clarification before implementation.

[fionaRust]

Tasks from Fiona. I've highlighted which ones I would like to review, and the rest could be reviewed by others if I'm not available.

• Remove hard coded geoid and instead take information from Iris cube: If you pass in the input cube you can access the coordinate system by doing coord_system = cube.coord("latitude").coord_system If it has the correct metadata I think it should return an iris.coord_systems.GeogCS instance ( https://scitools-iris.readthedocs.io/en/stable/generated/api/iris/coord_systems.html#iris.coord_systems.GeogCS) that should have attributes you can interrogate to find all the geiod information. You can always default to WGS84 parameters if you don't find any on the cube.
• Consider adding more unit tests at the function level. Fiona to review
• Understand the use case for cube_in_mask and the cube_in being different resolutions. If not necessary tidy up the code appropriately. Fiona to review/discuss
• Consider updating the bilinear acceptance test to show it gives the same answer as the old bilinear method.

[anja-bom]

Hi @fionaRust it looks like there is outstanding work here to do, is there anything you're waiting on BOM to do here?

[fionaRust]

@anja-bom we're not waiting for it and I think we realised a long time about that this was unlikely to get done. I've linked it to our own ticket for deciding whether to use these methods in our suite, which might include revisiting some of the things on this ticket. Hopefully this means it doesn't get forgotten if we decide to go further with it at some point.