feat: custom thinking animation with multi-file management[多文件管理的思考动画设置，头像大小可调]

[Andy365-365]

Summary

Allow users to customize the thinking animation shown during agent processing, with support for multiple animation files per profile.

Changes

Frontend

• Thinking animation modal in profile settings with a left preview / right thumbnail layout
• Multi-file upload — users can upload up to 5 animation files (GIF, WebP, MP4, WebM)
• Thumbnail list with individual delete buttons and click-to-select active animation
• Toggle to completely disable the thinking animation indicator
• i18n — full translations for both zh and en locales

Backend

• New API routes: PUT /api/hermes/profiles/:name/thinking-animation and DELETE /api/hermes/profiles/:name/thinking-animation
• Persistence: animation metadata saved as thinking-animation.json per profile
• Backward compatibility: legacy single-url format auto-migrated to items[] array
• Profile endpoints now return thinkingAnimation alongside avatar
• Body parser limit increased from 100KB to 50MB for large Base64 GIF uploads

Chat

• Thinking indicator reads custom animation from active profile instead of hardcoded GIF
• Respects enableThinking toggle to show/hide indicator

API Types

• New ProfileThinkingAnimation interface with items[] and activeIndex
• New API methods: updateProfileThinkingAnimation(), deleteProfileThinkingAnimation()
• Store exposes updateThinkingAnimation() and deleteThinkingAnimation()

Backward Compatibility

• url field retained — server auto-migrates old data to items[] format
• Profiles without custom animation use default light/dark GIF as before

Testing

• Tested with large GIF files (several MB) — no more 413 errors
• Verified profile endpoints return animation data correctly
• Verified toggle shows/hides indicator as expected

[Andy365-365]

@EKKOLearnAI

Hi ekko,

I've been using Hermes Web UI for a while and really like the project — great work! I'm new to open source contribution, so this is my first PR. Hope it's done right.

This PR adds two features:

1. Custom thinking animation — users can now upload and manage multiple animation files (GIF/WebP/MP4/WebM, up to 5) per profile, with a preview panel and toggle to enable/disable. This replaces the hardcoded thinking GIF with profile-level customization.

2. Configurable avatar size — avatar size is now adjustable per profile via a slider in the avatar modal (40-100px).

The PR has been rebased onto the latest main and the bodyParser conflict has been resolved (merged to 10mb, covering both avatar and animation upload needs).

Would be great to hear your feedback. Thanks!

[Andy365-365]

已与 main 解决冲突。

注意到 Build workflow 存在部分测试失败，在 main 分支上同样复现了这些失败，看起来与本 PR 的改动无关。Playwright e2e 检查已通过。

@EKKOLearnAI