fix(ai-moderation): implement prompt analysis recommendations

- Expand IMPHNEN domain rule to cover wildcard (*.imphnen.*)
- Trim redundant SARA examples from TEXT_ONLY_MODE (save ~950 tokens)
- Add debugging logs for channel culture injection into prompt
- Sync flag validation set with missing flags: potential_evasion, unclear_context

Co-Authored-By: Claude <noreply@anthropic.com>

scripts/truncate-all.sql

@@ -0,0 +1,22 @@
+-- Truncate all tables in correct order (respecting FK dependencies)
+-- attachments FK → messages (CASCADE), others are independent
+-- Use CASCADE to handle any remaining FK chains automatically
+
+TRUNCATE TABLE
+  messages,
+  attachments,
+  ui_state,
+  ai_analysis_runs,
+  voice_recordings,
+  user_reputations,
+  channel_cultures,
+  message_reviews,
+  moderation_actions,
+  retention_policies,
+  text_analysis_cache,
+  sticker_cache,
+  corrected_moderations,
+  user_profiles,
+  mascot_chat_messages,
+  muxer_jobs
+CASCADE;