Integration test: full export artifact generation pipeline

Integration tests require a dedicated CI Supabase project — provision one via Supabase Management API or use the CLI (supabase start for local Docker). Store project URL and service_role_key in GitHub Actions secrets mapped to environment variables read in test setup. The TestSupabaseHelper should use the service_role_key (bypassing RLS) for seeding and teardown, while the pipeline under test should use a test user's JWT to exercise real RLS policies. Use unique organization IDs per test run (UUID.v4()) to ensure parallel CI runs don't interfere.

Teardown should use ON DELETE CASCADE on seed records or explicit deletion in reverse dependency order. For the signed URL verification step, generate signed URLs with 2-hour expiry and use Dart's http package to assert HTTP 200 — this confirms the storage write actually succeeded. Document the test Supabase project setup in test/integration/README.md including required table schemas, RLS policies, and storage bucket configuration.

Integration tests using flutter_test with a real test Supabase instance. Use a TestSupabaseHelper class to manage seed data lifecycle: insertSeedActivities(), insertSeedAttachments(), tearDownTestData(). Each test should call setUp/tearDown via flutter_test hooks. Seed data constants defined in test/fixtures/bufdir_seed_data.dart.

Test scenarios: (1) canonical pipeline — 15 activities, 3 attachments, verify all 4 artifacts; (2) zero-activity organization — verify pipeline completes and produces valid empty-state artifacts; (3) large dataset smoke test — 200 activities, verify completion without timeout or error. Use http package to verify signed URLs return HTTP 200. Parse artifact contents using archive (ZIP), csv parser, and PDF byte inspection utilities from prior unit test tasks.