fix(automod): flow real LLM analysis + descriptive fallback

Root cause: ai-analysis-worker read llmResult.explanation and
llmResult.toxicityScore — fields the LLM pipeline never produces
(canonical AnalysisResult uses analysis/score). Every message fell back
to the bare template "Tidak ada indikasi pelanggaran." and the stored
score was always 0.

- Map analysis/score correctly; fallback now quotes the message content
- Prompt: ban generic analysis phrasing, require reply context
- LLM context: include replied-to message content (metadata.reference)
  so the model can explain what the user is replying to
- Frontend: show thread/channel names from metadata instead of raw IDs
  (message card, detail views, search overlay); detail panel now
  displays the ai_analysis text
- Auto-delete log/DM include the descriptive analysis as the reason

services/frontend/src/components/messages/search-overlay.tsx

@@ -4,6 +4,7 @@ import { Search, X } from "lucide-react";
 import { useEffect, useRef, useState } from "react";
 import { useQuery } from "@tanstack/react-query";
 import { messagesApi } from "@/lib/api";
+import { getMessageChannelLabel } from "@/lib/format";
 import type { MessageRecord } from "@/lib/types";
 
 interface SearchOverlayProps {
@@ -83,7 +84,7 @@ export function SearchOverlay({ open, onClose, onSelect }: SearchOverlayProps) {
               >
                 <div className="flex items-center gap-2 text-xs">
                   <span className="font-medium text-text-primary">{msg.username}</span>
-                  <span className="text-text-secondary/40">{msg.channel_id?.slice(0, 8)}</span>
+                  <span className="text-text-secondary/40">{getMessageChannelLabel(msg)}</span>
                 </div>
                 <p className="text-xs text-text-secondary/80 line-clamp-1 mt-0.5">{msg.content}</p>
               </button>