Example recipe using CORDEX data

[nhsavage]

replicate the content of examples/recipe_python.yml but using CORDEX data instead of CMIP data

• would include at least one rotated pole and one projected coordinate system model
• would also include more than one domain e.g. Europe and Africa

This would start to build confidence in the ability to use CORDEX data with ESMValTool

I have started to write this but come across problems which I will need to resolve before lodging this. Opening this issue so I can have a branch for the recipe I am trying to develop

[nhsavage]

this recipe contains two models and does the timeseries extraction as well as maps. The timeseries extraction requires ESMValCore#2177

• The area averaging in the python example does not work yet
• Multimodel statistics don't work either as the pre-processor attempts to merge them into a single cube and this fails because the two models have different coordinate systems despite being at a single point
• There are only models from Europe

[nhsavage]

note there are also some warnings:

 Can not guess bounds for coordinate lon from var lon: Multi-dimensional coordinate not supported: 'longitude'
 Can not guess bounds for coordinate lat from var lat: Multi-dimensional coordinate not supported: 'latitude'

At least for roated pole models, this should be possible to work around by finding the dimension coordinates and guessing bounds for those whatever they are.