更新版本号至2.1.0; 实现AI驱动的浏览器指纹分析 (#54)

<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Adds an AI-driven diagnosis panel to detect.html for browser fingerprint
analysis with OpenAI-compatible providers and an interactive chat.
Updates the extension to 2.1.0, with streaming stability, privacy, and
accessibility improvements.

- **New Features**
  - Diagnosis panel: multi-turn chat, message copy, Markdown export.
- Provider presets with per-provider local config (OpenAI, OpenRouter,
DeepSeek, Gemini, Qwen, etc.).
- Streaming chat and snapshot collector to feed structured detection
data; safe Markdown rendering; new en/zh i18n.

- **Refactors**
- Reworked detect.js to expose detection snapshots and session state;
added detect-ai.js; updated detect.html; shared modules: ai-chat-client,
ai-provider-presets, copy-button.
- Extracted provider select population into AIProviderPresetUtils;
improved streaming buffer handling and [DONE] processing; switched to
ResourceManager.setTimeout; reduced config writes by using change events
only.
- Privacy safeguards and accessibility improvements in the diagnosis
panel.
- Docs updated with AI diagnosis notes and project structure; README
streamlined.

<sup>Written for commit 8a22fcc.
Summary will update on new commits.</sup>

<!-- End of auto-generated description by cubic. -->

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
- AI Diagnosis panel with multiple provider presets, per-provider local
config, API key handling, and real-time streamed responses
- Interactive multi-turn chat with follow-ups, start/stop/clear
controls, and abort support
  - Export conversations to Markdown and copy individual messages

* **Improvements**
- Refactored detection flow with refresh control and a stable detection
snapshot used by AI diagnosis
  - Full English and Chinese UI text for the AI workflow

* **Chores**
  - Version bumped to 2.1.0
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Author: ChuwuYo

README.md

@@ -82,6 +82,7 @@ Features
   - Internationalization API (Intl) information
   - WebRTC local IP leakage
   - Canvas, WebGL, and AudioContext fingerprinting information
+  - AI Diagnosis panel with OpenAI-compatible providers, multi-turn chat, per-provider local config, Markdown export, and message copy actions
 * **Debugging Tools**: `debug.html` provides debugging and diagnostic features, including:
   - View `declarativeNetRequest` dynamic rule details
   - Multi-endpoint request header testing
@@ -131,6 +132,17 @@ Usage
 3.  **Apply Settings**: Click the "Apply Changes" button at the bottom of the interface. Your settings will be saved and immediately applied to new network requests.
 4.  **Verification and Debugging**: Click the "Detection Page" or "Debugging Tools" links provided in the popup to verify if the language settings are effective, or to diagnose issues when they occur.
 
+### AI Diagnosis Notes
+
+The `detect.html` page now includes an AI diagnosis panel. It sends the current structured detection snapshot to an OpenAI-compatible model and turns the raw data into user-friendly explanations.
+
+Current implementation notes:
+
+* Querying an external AI provider sends a structured detection snapshot to that provider. This can include fingerprint-related signals such as language settings, locale and timezone data, partial browser fingerprint fields, and redacted WebRTC or header-derived context. Treat AI diagnosis as third-party data sharing and only use providers you trust.
+* AI configuration is stored locally in `chrome.storage.local`, grouped by provider
+* Replies follow the current detect page language by default
+* Exported Markdown only contains visible conversation, excluding the hidden system prompt and hidden initial detection snapshot injection messages
+
 ***
 
 <div align="center">