KGO test with altitude factory

# 🐜 Bug Report

None of our rose stem tests are currently testing behaviour with an altitude derived coordinate. We should be testing that we can load, save and regrid data with [complete hybrid height coordinates](https://cfconventions.org/Data/cf-conventions/cf-conventions-1.9/cf-conventions.html#atmosphere-hybrid-height-coordinate). Note the following mapping from the CF equation terms to iris terms:

| CF | Iris |
|--|--|
| z | altitude |
| a | level height |
| b | sigma |
| orog | surface altitude |

We do have a 3D to 3D regrid, but since there's no `surface altitude` in the source files, there's no derived altitude coordinate.

We should add a synthetic `surface altitude` to the source cube to confirm that the behaviour when a derived altitude is present is as expected.

From our unittesting and real world usage, I think ANTS behaviour is correct - this is plugging a gap in the functional tests.

## Version
This bug exists in ANTS/head of main

Start with the existing rose stem tests and their 3D-to-3D regrid setup. Add a synthetic surface altitude to the source cube, using the stated CF-to-Iris term mapping, and verify that data can be loaded, saved, and regridded with the derived altitude coordinate present.

Written by the indexing model from the issue text.