feat: BeautyLeads B2B cosmetics frontend on port 7788

New service (app/beauty_main.py) sharing the same /data volume:
- Separate FastAPI app running on port 7788
- beauty_ai.py: brand universe scan (~650 brands), portfolio match
  detection against OUR_BRANDS, Gemini B2B assessment prompt in Spanish
  returning quality/categories/dist_matches/outreach_email
- beauty_queue table + beauty_lead_quality/beauty_assessment columns
  in enriched_domains (with migrations)
- Endpoints: /api/beauty/assess/batch, /api/beauty/leads,
  /api/beauty/status, /api/beauty/export, /api/beauty/reset
- Static frontend: Browse (beauty/ecommerce pre-filtered, no CMS/SSL/KD
  columns), Validator, B2B Pipeline (brand chips, expandable outreach),
  Pre-screen, Export CSV
- docker-compose: second 'beauty' service with shared data volume
- Dockerfile: expose 7788 alongside 6677

Also: add 'error' prescreen_status handling + UI (orange stat box,
filter option) for 4xx/5xx HTTP responses

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

Dockerfile

@@ -11,6 +11,6 @@ RUN pip install --no-cache-dir -r requirements.txt
 
 COPY app/ ./app/
 
-EXPOSE 6677
+EXPOSE 6677 7788
 
 CMD ["python", "-m", "app.main"]