fix: hide framework labels in generated content + editable hashtags

- Add explicit instruction to LLM: never write framework labels (PROBLEMA,
  AGITAZIONE, SOLUZIONE, etc.) — use them as invisible narrative structure only
- Replace static hashtag chips with HashtagEditor component:
  - Click hashtag to edit inline
  - Click X to remove
  - Input field to add new hashtags
  - Save button to persist changes to DB

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

backend/app/services/content.py

@@ -41,7 +41,12 @@ def generate_post_text(
         f"Your expertise covers: {topics_str}. "
         f"Your communication style is {tone}. "
         f"You create authentic, engaging content that resonates with your audience. "
-        f"Never reveal you are an AI. Write as {name} would naturally write."
+        f"Never reveal you are an AI. Write as {name} would naturally write.\n\n"
+        f"REGOLA CRITICA: Se ti viene indicata una tecnica narrativa (PAS, AIDA, Storytelling, ecc.), "
+        f"usala SOLO come struttura invisibile del testo. "
+        f"NON scrivere MAI le etichette del framework nel post (es. non scrivere 'PROBLEMA:', "
+        f"'AGITAZIONE:', 'SOLUZIONE:', 'ATTENZIONE:', 'INTERESSE:', ecc.). "
+        f"Il lettore non deve percepire alcun framework — deve sembrare un post naturale e spontaneo."
     )
 
     # Platform-specific instructions