DIG-1344: make sure no dup IDs in schemas

[daisieh]

Validate that there are no duplicate IDs within an ingest dataset, at least within each schema.

There are tests to check for this in the test file.

NOTE: there are also extra bonus fixes here:

• when something fails during CSVConvert, it will write out the current state of INDEXED_DATA to x_indexed.json, including all calculated fields in a virtual sheet called CALCULATED
• added a mapping function that just concatenates all the values passed to it (seemed possibly useful)
• added a bunch of extra checks in the validation methods to check for keys that exist but are null. (this is so that you can validate the output of katsu's authorized/donor_with_clinical_data/ endpoint and have that pass)

None of these bonus fixes should affect behavior, at least as far as I can tell.

[daisieh]

I'm pulling this back to draft because I keep pushing changes onto the branch and Javi hasn't had time to review anyway.

[daisieh]

Rebased this branch to remove the jsonschema stuff. That will go into a different PR.