feat: replace AI matting with chroma-based checkerboard detection

The transparency-fixer now directly detects the baked-in checkerboard
pattern using per-pixel chroma analysis instead of BiRefNet AI matting.
Achromatic pixels in the gray range are classified as background
(transparent), chromatic pixels as foreground (opaque), with smooth
transitions at anti-aliased edges.

- No longer requires Python sidecar or background-removal bundle
- Watermark removal uses Sharp median(5) filter pre-processing
- Moved tool from "ai" to "utilities" category
- Removed from PYTHON_SIDECAR_TOOLS and background-removal enablesTools
- Near-instant processing (pure Sharp, no model inference)

packages/shared/src/constants.ts

@@ -250,18 +250,18 @@ export const TOOLS: Tool[] = [
     route: "/content-aware-resize",
   },
   {
-    id: "content-aware-crop",
-    name: "Content-Aware Crop",
-    description: "Extend canvas beyond image bounds with AI-powered fill",
+    id: "ai-canvas-expand",
+    name: "AI Canvas Expand",
+    description: "Expand canvas beyond image bounds with AI-powered fill",
     category: "ai",
     icon: "Expand",
-    route: "/content-aware-crop",
+    route: "/ai-canvas-expand",
   },
   {
     id: "transparency-fixer",
     name: "PNG Transparency Fixer",
     description: "Fix fake transparent PNGs in one click",
-    category: "ai",
+    category: "utilities",
     icon: "ShieldCheck",
     route: "/transparency-fixer",
   },
@@ -1247,6 +1247,5 @@ export const PYTHON_SIDECAR_TOOLS = [
   "red-eye-removal",
   "restore-photo",
   "passport-photo",
-  "transparency-fixer",
-  "content-aware-crop",
+  "ai-canvas-expand",
 ] as const;