Data hosting

[gabrieletijunaityte]

For now, it is expected that developers download model_ready csv and caption template JSON data from the shared folders on Teams.

Once the project data is finalised and published, we should host it externally (e.g. zenodo) and implement Pooch retrieval for the csv and JSON data, also, potentially, for the relevant EO modalities.

[vdplasthijs]

Copying text from #22 which essentially a sub-issue of this:

The repo contains some example data for the butterfly use case in data/s2bms/.

When someone starts using the repo, they can either 1) use local/aether/data/ as their data folder (which is in .gitignore to prevent syncing data) or 2) set an external data_dir via .env or config/paths/local.yaml.

In the second case; currently the existing example data is not automatically copied to the new data_dir, leading to an error in BaseDataSet. (Currently this can be resolved by manually copying data/s2bms/ into data_dir.

Best way to resolve this would be to use pooch to automatically download & sync the latest example data from the github repo to the local data_dir.