OMOP schema "stretches", issues

while loading into DuckDB for DQ, I found some OMOP schema violations we'll need to work on

[ ] OMOP wants integer IDs, does AoU live up to that? Will need to tweak the hash solution for IDs, may break other existing IDs we were hoping to use, and may still, but in an added _source_id column or somesuch.
[ ] The CCDA date values don't have dashes. DuckDB and parsing code in Python want real ISO 8601 YYYY-MM-dd style. I have some simple fixes in place, but want to verify what we should be able to consistently expect from CCDA documents. If they are rock-solid YYYYMMDD, we're good. Otherwise (surely) we'll need some more sophisticated date parsing.
[ ] so far in Person, the demo columns for race, gender and ethnicity are NON NULL in OMOP, but we don't always get them in our test data.

The first and last here are relaxed in the OMOP/DuckDB code here and need to be addressed as we integrate into the wider world.

cladteam / CCDA_OMOP_by_Python