Sample data

[wtbarnes]

Both for testing and examples, it's going to be really convenient to have some sample data. I'm thinking that we could have both "real" data and synthetic data, where the "real" data are based off of canonical DEMs from ARs, QS, CH, etc. (like in CHIANTI). This has some interesting overlap with @ehsteve's solar_datasets repo.

We could also have a few "ground truth" DEM distributions (e.g. just a single Gaussian) and then some fake response functions from an idealized telescope to generate synthetic data. Having these would be really useful for easily testing out methods as we add them.

We could then have convenience functions that generate NDCube/NDCollections for this sample data.

[ehsteve]

My idea for solar_datasets is that just like sunpy provides solar constants and affiliated packages should import sunpy and use them, affiliated packages would do the same for solar_datasets.