reform hh parsing pipeline

[chunhochow]

For 2022/, it seems like only the raw hh file is used in all of the summary notebooks in 04b-summary_notebooks (via load_hh_raw() in utils). If this is indeed the case (to be verified) and the processed hh files aren't used, should we produce so many iterations/versions of the hh file as it goes through the different steps of the processing pipeline? This seems to just be adding to the cognitive load and confusion of what exactly the files are after each specific step of the pipeline.