Document how the synthetic data for the scaling tests is generated

[gcapes]

@jburba's comments, copied from a previous issue

Regarding generating the scaling data, I made the scaling data in a jupyter notebook and the process is a little rough around the edges. The process could probably be made into a script, but that would require some effort. To get all the metadata right, you have to generate the csv and yaml files used to make an empty data container. Formatting the csv file via python is easy enough. Formatting the yaml via python is a little trickier in my experience, specifically in regards to typing and getting ints/floats/strings to be formatted properly in the output yaml file.

Regarding reproducibility, the scaling data generation is reproducible as each baseline gets a copy of the same data (from the test data already distributed with hydra-pspec).