Add more configurable parameters to dataset methods?

[geek-yang]

In the methods of dataset classes, there are many parameters which can be made configurable, for instance, chunks for loading data:

zampy/src/zampy/datasets/ecmwf_dataset.py

Line 123 in fa72728

ds = xr.open_mfdataset(files, chunks={"latitude": 200, "longitude": 200})

These configurable data could be added as arguments to the methods in each dataset class. However, since currently we want our users to use zampy with recipes only (https://github.com/EcoExtreML/zampy/blob/main/docs/using_zampy.md), this may not help too much. We need to think about the design to bridge the gap.

[BSchilperoort]

However, since currently we want our users to use zampy with recipes only

Well, there is the config file too. ESMValTool has a dask-specific config file.

The recipes determine what data is downloaded and how it's converted. However, things like the configuration of dask, login keys, etc. can be in a user config.