feat(build): 全面优化打包流程，Docker 镜像自动发布到 GHCR

Docker 优化：
- Dockerfile 层缓存（requirements/lockfile 先复制再安装）
- ARG 可配置镜像源，国际用户可覆盖为默认源
- 前端 Dockerfile 改用 corepack + frozen-lockfile
- 精简 .dockerignore，排除 .git 和 Tauri 构建产物

CI/CD 优化：
- docker-build 自动推送到 GHCR，支持 amd64/arm64 双架构
- 桌面端 CI 增加 pip/pnpm/cargo 缓存，升级 actions 版本
- Python 版本对齐为 3.11，增加 Linux 构建矩阵
- build.sh 加 -y 覆盖标志

文档更新：
- README Docker 部署简化为 docker pull + docker run

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

backend/Dockerfile.gpu

@@ -1,16 +1,27 @@
 FROM nvidia/cuda:12.4.1-cudnn-runtime-ubuntu22.04
 
-RUN apt update && \
-    apt install -y ffmpeg python3-pip && \
-    apt clean all && \
+ARG APT_MIRROR=mirrors.tuna.tsinghua.edu.cn
+ARG PIP_INDEX=https://pypi.tuna.tsinghua.edu.cn/simple
+
+RUN rm -f /etc/apt/sources.list && \
+    rm -rf /etc/apt/sources.list.d/* && \
+    echo "deb https://${APT_MIRROR}/ubuntu jammy main restricted universe multiverse" > /etc/apt/sources.list && \
+    echo "deb https://${APT_MIRROR}/ubuntu jammy-updates main restricted universe multiverse" >> /etc/apt/sources.list && \
+    echo "deb https://${APT_MIRROR}/ubuntu jammy-security main restricted universe multiverse" >> /etc/apt/sources.list && \
+    apt-get update && \
+    apt-get install -y --no-install-recommends ffmpeg python3-pip curl && \
     rm -rf /var/lib/apt/lists/*
 
-# 设置 Hugging Face 镜像源环境变量
 ENV HF_ENDPOINT=https://hf-mirror.com
 
 WORKDIR /app
+
+# 先复制 requirements.txt 利用层缓存
+COPY ./backend/requirements.txt /app/requirements.txt
+RUN pip install --no-cache-dir -i ${PIP_INDEX} -r requirements.txt && \
+    pip install --no-cache-dir -i ${PIP_INDEX} 'transformers[torch]>=4.23'
+
+# 再复制应用代码
 COPY ./backend /app
-RUN pip install --no-cache-dir -i https://pypi.mirrors.ustc.edu.cn/simple -r requirements.txt
-RUN pip install --no-cache-dir -i https://pypi.mirrors.ustc.edu.cn/simple 'transformers[torch]>=4.23'
 
 CMD ["python3", "main.py"]