Design a normalized SQL schema with indexes and constraints

Design a SQL schema for the following data, for [PostgreSQL / MySQL / SQLite — specify], to be used with [ORM name, e.g. Prisma/Drizzle/SQLAlchemy, or "raw SQL"].

Data description: [describe your entities and how they relate in plain language, e.g. "users belong to workspaces, workspaces have projects, projects have tasks assigned to users, and tasks can have comments"]

Known query patterns: [list the main ways this data will be queried, e.g. "list all tasks for a workspace sorted by due date", "find all comments by a user across projects"]

Expected scale: [rough row counts if known, e.g. "thousands of workspaces, tens of thousands of tasks" — or say "unknown, design conservatively"]

Please:
1. Propose a normalized schema (tables, columns, types) — call out any deliberate denormalization and why
2. Define primary keys, foreign keys, and ON DELETE behavior (cascade, restrict, set null) for each relationship, with reasoning
3. Add NOT NULL, UNIQUE, and CHECK constraints wherever the data model implies them, rather than relying on application-level validation alone
4. Recommend indexes based on the query patterns I listed, and explain which column(s) each index supports
5. Point out any many-to-many relationships and design the join table properly
6. Write this as an actual migration file (or ORM schema definition) in the format my stack uses
7. Include a couple of example queries for the query patterns I listed, using the schema you designed

Flag anything where you had to guess at a business rule (e.g. "can a task exist without a project?") instead of assuming — list these as open questions.