Speed up downloading of Cubico data to make it more accessible

[charlie9578]

It takes a long time to download all the Cubico data, and also corresponding reanalysis data, and the file sizes are large! This makes the dataset hard for people to use.

I'm thinking it would be good to provide the required files to run examples (and some additional signals to explore) to make the data more accessible. Initially I'm thinking just to use the "to_csv" function, upload the data to Zenodo, use the download tool, then load in the project data, which should signficantly speed things up and make the data more accessible.

The project_Cubico script should still contain the full functionality though for reference, or even converted into a notebook, as others will want to see how they could use their own data.

Any thoughts?

[RHammond2]

@charlie9578 if the data were able to be readily available on Zenodo, then I think that would speed up the example quite a bit. I think we've been hesitant to add more data to the repository itself. That said, if 20 year profiles were made available from the get-go that would speed up much of the work of the example.

A couple of suggestions on how this could be set up to be a little quicker:

• Set the start_date and end_date for the reanalysis periods
• Add the data to Zenodo, then it's a one step download process, though not much faster
• The current suite of data is ~7GB, so it could be helpful to remove any intermediary files after the download process (mostly reanalysis data), and reassess how much extra data it would ultimately add to the repository. I think the cleanup steps are still important to keep available as well, so it might be good to have a more formal chat about this (and likely the other open issues too).