feat(canvas): resolution/ratio composer settings and quality/count controls

Replace the ratio-baked size presets with a resolution tier (1K/2K/4K) plus
ratio selection and normalized dimensions, send imageQuality/imageCount to
the agent, and add the output_format field to GeneratorTaskInputArgs.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

frontend/src/i18n/locales/en-US.json

@@ -251,6 +251,7 @@
   "qualityHigh": "High",
   "qualityMedium": "Medium",
   "qualityLow": "Low",
+  "imageResolutionLabel": "Base resolution",
   "imageSizeLabel": "Size",
   "imageModeDirective": "Image generation settings: quality {quality}, size {width}x{height}, ratio {ratio}, count {count}. Pure image mode: generate the image directly from the user's input without extra chat.",
   "quickEdit": "Quick edit",

frontend/src/i18n/locales/zh-CN.json

@@ -251,6 +251,7 @@
   "qualityHigh": "高",
   "qualityMedium": "中",
   "qualityLow": "低",
+  "imageResolutionLabel": "基准分辨率",
   "imageSizeLabel": "尺寸",
   "imageModeDirective": "图片生成参数：质量 {quality}，尺寸 {width}x{height}，比例 {ratio}，数量 {count} 张。纯生图模式，请直接围绕用户输入生成图片，不展开额外聊天。",
   "quickEdit": "快捷编辑",