feat: implement AI conversation generation with Mistral

- Install Mistral AI SDK and required dependencies
- Create API endpoint for generating podcast conversations using structured data
- Generate conversations between two hosts with distinct personalities:
  * Host 1: Bubbly, excited, and enthusiastic
  * Host 2: Skeptical, sarcastic, and thoughtful
- Include emotional expressions in brackets like [giggles], [sarcastically]
- Use same language as scraped content (English, Italian, etc.)
- Update frontend to automatically generate conversation after scraping
- Remove mock data and unused variables
- Add proper error handling and loading states

🤖 Generated with [Claude Code](https://claude.ai/code)

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

package.json

@@ -9,16 +9,19 @@
     "lint": "eslint"
   },
   "dependencies": {
+    "@ai-sdk/mistral": "^2.0.15",
     "@mendable/firecrawl-js": "^4.3.5",
     "@radix-ui/react-progress": "^1.1.7",
     "@radix-ui/react-slot": "^1.2.3",
+    "ai": "^5.0.48",
     "class-variance-authority": "^0.7.1",
     "clsx": "^2.1.1",
     "lucide-react": "^0.544.0",
     "next": "15.5.3",
     "react": "19.1.0",
     "react-dom": "19.1.0",
-    "tailwind-merge": "^3.3.1"
+    "tailwind-merge": "^3.3.1",
+    "zod": "^3.25.76"
   },
   "devDependencies": {
     "@eslint/eslintrc": "^3",