Make semantically required nullable DB columns NOT NULL

Several DB columns are semantically required but are declared nullable in the current schema (e.g. users.created_at, users.modified_at). This epic tracks auditing the schema for such columns and migrating them to NOT NULL.

Scope:

- Audit all tables for columns that must always have a value but are declared nullable (audit timestamps such as created_at / modified_at are the primary examples).
- For each column: clean up or backfill existing NULL rows, add a server-side default where appropriate, and add an Alembic migration setting NOT NULL.
- Update the corresponding ORM models and type annotations.

Child issues will be created per table/column group.

JIRA Issue: BA-7215

Start by auditing the schema, ORM models, and Alembic migrations for semantically required nullable columns, beginning with users.created_at and users.modified_at. Check for existing NULL rows and determine appropriate server-side defaults. Done means affected rows are handled, models and type annotations are updated, and each column has a migration enforcing NOT NULL.

Written by the indexing model from the issue text.