feat: v2.8 — Instagram Reels source + TikTok ScrapeCreators migration

Add Instagram Reels as the 8th research source via ScrapeCreators API.
One API key (SCRAPECREATORS_API_KEY) now covers both TikTok and Instagram.

- Add scripts/lib/instagram.py: keyword search, transcript extraction,
  relevance scoring, engagement metrics (views, likes, comments)
- Add InstagramItem to schema, normalization, scoring, dedup, rendering
- Add Instagram to orchestrator pipeline, watchlist, and UI spinners
- Update SKILL.md: stats template, citation priority, item format,
  URL-to-name extraction rules, anti-Sources instruction
- Update README and CHANGELOG for v2.8
- Fix: Instagram/TikTok not running in --search= web-only path
- Fix: web stats line showing full URLs instead of domain names
- Replace APIFY_API_TOKEN with SCRAPECREATORS_API_KEY throughout

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

scripts/watchlist.py

@@ -211,6 +211,16 @@ def _run_topic(topic: dict) -> dict:
                 "engagement_score": (item.get("engagement") or {}).get("views", 0),
                 "relevance_score": item.get("relevance", 0),
             })
+        for item in data.get("instagram", []):
+            findings.append({
+                "source": "instagram",
+                "url": item.get("url", ""),
+                "title": (item.get("caption_snippet", "") or "")[:120],
+                "author": item.get("author_name", ""),
+                "content": item.get("caption_snippet", ""),
+                "engagement_score": (item.get("engagement") or {}).get("views", 0),
+                "relevance_score": item.get("relevance", 0),
+            })
 
         # Store with dedup
         counts = store.store_findings(run_id, topic_id, findings)