This is unfortunately not posisble with sh-py. And it is also not that easy as the archive data that are available in the cloud, are in the old .SAFE format and the current version of Sen2Cor does not work with it (nicely, at least). This is also the reason, why we have not processed them prior to 2017.
If you want to use them, you will have to use sen2cor software “manually”. After you do it, you can ingest them to Sentinel Hub (using “Bring your own data” feature) and use them in the same way as the rest.
Note, however, that there is a decent chance that ESA will do this in the next 6-12 months. There is a “large reprocessing campaign” planned to start this year, where they will reprocess all L1C data with the new GRI. We assume that they will then also reprocess/create L2A data. So, if everything goes as planned, you will have access to these products (in even better quality than now) in not too distant future.