Open Dungeon 0.4.1: ComfyUI backend

Bring your own ComfyUI, and a tidier repo.

• ComfyUI is now a built-in image backend. Already run ComfyUI? Pick ComfyUI in the Images panel, confirm the server URL (default http://127.0.0.1:8188), and choose a checkpoint — the picker lists your server's own checkpoints, or leave it on Auto. The app drives ComfyUI directly over its HTTP API: no FLUX worker, no extra setup, works on any hardware ComfyUI supports (including AMD cards outside the native ROCm list). Sizes clamp to 1024/1344 long side for SDXL-class checkpoints; character portrait references don't apply to this backend.
• Cleaner repo layout. The secondary Windows tools (Stop-Windows.bat, Diagnose-Windows.bat, the CPU launcher, and the image smoke/loop tests) moved into the windows\ folder. Launch-Windows.bat and Launch.command stay at the root and remain one-click. A dead day-one image worker was removed.
• README rewritten around the essentials, with the deep material split into focused guides: text backends, image generation, Windows, and configuration.

Install:

• macOS Apple Silicon: open the DMG, drag Open Dungeon to Applications, and open it (right-click then Open the first time, since it is unsigned).
• Windows: download Open-Dungeon-0.4.1-Windows.zip, unzip it, then run Launch-Windows.bat.

Validation: Windows smoke passed on the release commit, local lint/build passed, the ComfyUI backend verified end-to-end against a protocol-accurate mock server (generation, checkpoint listing, and error paths), and the DMG verified with hdiutil verify.

Full changelog: v0.4.0...v0.4.1