Prototype delayed dask dataframes for better computation scaling #9
Description
Currently all computation is multithreaded by default and that doesn't leave a lot of room to the user in terms of deciding how and when they want to actually gather data or compute a result.
The datasets module can easily be switched over to dask for data gathering and caching
The keywords module should be looked at for how to use dask dataframes for just-in-time compute and out-of-memory compute.