Implement N+1-safe cross-chapter JOIN query

PostgREST foreign key embedding syntax for a join: supabase.from('contact_chapters').select('chapter_id, chapters(name), activities(id, activity_type, activity_date, duration_minutes, summary, is_billable)').eq('contact_id', contactId).order('activity_date', referencedTable: 'activities', ascending: false). Note that PostgREST returns a nested JSON structure; you will need to flatten it: for each contact_chapter row, extract chapter_id and chapters.name, then iterate the embedded activities array. This is still a single HTTP call. Alternatively, use a Supabase RPC (stored procedure) for more control over the SQL.

The RPC approach is preferable if the join involves more than 2 levels of nesting or if you need DISTINCT ON for deduplication. Create the RPC in a migration file and call it via supabase.rpc('get_cross_chapter_activities', params: {'p_contact_id': contactId}). Ensure the function is defined with SECURITY INVOKER (not DEFINER) so it runs under the caller's RLS context.

Write unit tests using a mocked SupabaseClient: (1) mock a successful response with 3 activities across 2 chapters and verify the result list contains 3 CrossChapterActivityResult objects with correct chapterName values; (2) mock an empty response and verify an empty list is returned; (3) mock a PostgrestException and verify ActivityQueryException is thrown. Write integration tests against a local Supabase instance: seed contact_chapters with 3 chapter affiliations, seed activities for each chapter, call fetchActivitiesForContact, and assert all activities are returned in date-descending order; verify that activities from a 4th chapter (not affiliated) are not returned. Use flutter_test for all tests.