feat: 屏幕截图组件实现图像来源与截图范围（原 P2 问题）

- vision.ts 重写 Image_Screenshot：pictureSource=1 当前屏幕(desktopCapturer)/2 指定图片文件(image_path 作输入, 输出走输出目录+文件名, 未配置则原样返回)；shotRange=2 框选区域按 position/action 区域 JSON 裁剪(pngjs bitblt)，区域支持 {x,y,width,height}/[x,y,w,h]/left,top/角点写法并夹取边界
- 移除 browser.ts 重复注册，实现单点维护
- catalog: pictureSource/shotRange 补下拉选项(当前屏幕/指定图片文件, 整个屏幕/框选区域)
- typecheck 通过；browser/cv-vision/capability-audit/app-system 回归 50 项全过

src/main/capabilities/browser.ts

@@ -178,21 +178,4 @@ register("WebBrowser_Screenshot", async (ctx, node, args) => {
   return file;
 });
 
-register("Image_Screenshot", async (ctx, node, args) => {
-  const { desktopCapturer, screen } = await import("electron");
-  const folder = String(args.outImagePath ?? args.image_path ?? ctx.evidenceDir);
-  const name = String(args.outImageName ?? `${node.id}.png`);
-  const { promises: fs } = await import("node:fs");
-  const path = await import("node:path");
-  const target = path.resolve(folder.endsWith(".png") ? folder : path.join(folder, name));
-  const primary = screen.getPrimaryDisplay();
-  const sources = await desktopCapturer.getSources({
-    types: ["screen"],
-    thumbnailSize: { width: primary.size.width, height: primary.size.height }
-  });
-  const source = sources[0];
-  if (!source) throw new RuntimeError("desktop_error", "无法获取屏幕截图");
-  await fs.mkdir(path.dirname(target), { recursive: true });
-  await fs.writeFile(target, source.thumbnail.toPNG());
-  return target;
-});
+// Image_Screenshot 统一在 vision.ts 实现（屏幕/文件来源 + 区域裁剪），此处不再重复注册