Feature: Add LLM COT Rendering support for WebUI

[danielaskdd]

Add LLM COT Rendering support for WebUI

📋 Summary

This PR introduces a new feature that visualizes AI thinking processes in the chat interface, providing users with transparency into how the AI formulates responses. The feature includes real-time thinking content display, timing statistics, and an intuitive expandable UI.

✨ Features Added

🎯 Core Functionality

• Real-time Thinking Visualization: Parse and display AI thinking content using <think> tags during streaming responses
• Thinking Time Tracking: Accurate measurement and display of thinking duration with millisecond precision
• Expandable UI: Collapsible thinking sections with smooth animations and intuitive controls
• Streaming Support: Real-time parsing of thinking content as responses are received

🌐 User Experience Improvements

• Visual Indicators: Clear thinking status with loading animations and chevron icons
• Content Separation: Clean separation between thinking process and final response
• Interactive Controls: Click-to-expand functionality for thinking content
• Responsive Design: Optimized layout for different screen sizes

🌍 Internationalization

• Multi-language Support: Added translations for 5 languages (EN, ZH, ZH_TW, AR, FR)
• Localized Timing Display: Culture-appropriate time formatting
• Accessible Labels: Screen reader friendly interface elements