The draft import / ingestion process has fallen behind changes to the other data models. The import_submission() view, Submission serializer, and perhaps other related views need to be updated to work with the additional and restructured fields that have come in recently. This may be more than what came in with #790, but that's what I noticed.
Possibly related to #726
As a possibly helpful hint, decorating the Submission.from_rpcapi_draft() method like the following reveals some immediate issues, depending on whether your syntax highlighter knows about the datatracker rpcapi client models.
def from_rpcapi_draft(cls, draft: rpcapi_client.FullDraft):