Performance test activity query with large datasets

The primary performance bottleneck is almost certainly the join between activities, peer_mentors, and chapters tables filtered by organization_id. Indexes to consider: (1) CREATE INDEX idx_activities_org_period ON activities(organization_id, reporting_period_id); (2) CREATE INDEX idx_activities_peer_mentor ON activities(peer_mentor_id); (3) CREATE INDEX idx_peer_mentors_org ON peer_mentors(organization_id). Deduplication (removing double-counted activities where a peer mentor belongs to multiple chapters) should be a pure Dart operation on the fetched list — measure this separately from the DB query. Use Supabase's rpc() to call a stored procedure for EXPLAIN ANALYZE output, or connect via a raw pg connection in test-only code.

Seed data generation: use a DataFactory helper that generates realistic Norwegian activity records with proper date ranges within the reporting period. Commit index migrations as numbered SQL files in supabase/migrations/ following the project's existing migration naming convention.

Performance tests using flutter_test with Stopwatch for timing measurement. Structure as: (1) one-time setUp that inserts 1000 synthetic activities via batch Supabase inserts (use upsert in chunks of 100 to stay within rate limits); (2) timed measurement block using Stopwatch.start/stop around the BufdirActivityQueryService.fetchAllForOrganization() call; (3) assertion that elapsed milliseconds < 5000; (4) separate timed block for deduplication logic. Log Stopwatch results with print() for CI visibility. After initial run, run EXPLAIN ANALYZE via a raw SQL query through Supabase service role connection and log the plan output.

Apply identified indexes (e.g., on organization_id, reporting_period, peer_mentor_id) in a new migration file. Re-run and assert improvement. Test teardown deletes all seeded records.