Preprocess without Dask

[dthain]

Per @hannahbnelson Create a preprocessing function for DDR (and/or Coffee) that doesn't make use of Dask.

[hannahbnelson]

For a bit more context, in my "processor/executor" coffea setup, I use something like this

from coffea import processor
exec_instance = processor.TaskVineExecutor(**executor_args)
runner = processor.Runner(
                exec_instance,
                schema=NanoAODSchema,
                chunksize=chunksize,
                maxchunks=nchunks,
                xrootdtimeout=300,
            )
output = runner.run(fileset=flist, processor_instance=proc_instance, treename=treename)

Where fileset is a dictionary of this structure: fileset = {'dataset': [file1, file2],}.
The coffea Runner class has a preprocessing function that uses this input format. The call to that is here, and Runner.preprocess() is here