Add local GPU inference: auto-detect GPU, auto-download best diffusion models

Adds AI_PROVIDER=local_gpu — a fully self-contained GPU inference engine
using HuggingFace Diffusers that requires zero InvokeAI/ComfyUI setup.
All existing providers (InvokeAI, ComfyUI, OpenAI, Replicate) remain intact
and can be mixed with local GPU via per-operation overrides.

New features:
- GPU auto-detection (CUDA/NVIDIA, MPS/Apple Silicon, CPU fallback)
- VRAM-tiered model selection:
    ultra ≥16 GB → SDXL inpaint + SDXL base
    high  8-16 GB → SDXL inpaint + SDXL base
    medium 4-8 GB → SD 2.x inpaint + SD 2.1
    low  <4 GB   → SD 2.x (small)
- Auto-download model weights to HuggingFace disk cache at startup
  (background task; first request loads from local disk, not internet)
- LRU pipeline cache evicts oldest GPU pipeline when VRAM limit reached
- Per-operation model overrides via HF_MODEL_INPAINT / HF_MODEL_TXT2IMG etc.
- Optional HF_TOKEN for gated/private HuggingFace models

New files:
- backend/app/services/gpu_detect.py   — GPU detection + tier/model mapping
- backend/app/services/local_diffusion.py — Diffusers provider + LRU cache
- backend/app/routers/gpu_status.py    — GET /api/gpu/status, POST /api/gpu/prefetch
- backend/requirements.gpu.txt         — Diffusers ecosystem deps (GPU only)
- docker-compose.gpu.yml               — NVIDIA GPU compose (one-command startup)
- Dockerfile.gpu                       — pytorch/pytorch:2.1.0-cuda12.1 base image
- scripts/gpu_setup.py                 — Startup GPU info logger

Modified:
- backend/app/config.py                — local_gpu settings added
- backend/app/services/remote_provider.py — local_gpu registered as provider
- backend/app/routers/ai_tools.py      — /api/config exposes GPU tier + caps
- backend/app/main.py                  — GPU router + background prefetch task
- backend/entrypoint.sh                — runs gpu_setup.py at container start
- .env.example                         — local_gpu documented as first option

Quick start with GPU:
  docker compose -f docker-compose.gpu.yml up --build

https://claude.ai/code/session_01WVDg7amsy1TTtxvpku7bcM

.env.example

@@ -14,18 +14,37 @@
 # =============================================================================
 # STEP 1: Choose AI Provider
 # =============================================================================
-# Options: mock, openai, stability, replicate
+# Options: local_gpu, mock, openai, stability, replicate, invokeai, comfyui
 #
-#   mock      = Free, but returns original image unchanged (for testing UI)
-#   openai    = DALL-E 2 inpainting (~$0.02/image) - lower quality
-#   stability = Stability AI SDXL (~$0.01/image) - good quality
-#   replicate = Multiple models (~$0.002-0.03/image) - RECOMMENDED
+#   local_gpu = FREE, runs on YOUR GPU — best option if you have an NVIDIA card
+#               (use docker-compose.gpu.yml — models auto-download on first use)
+#   mock      = Free, returns original image unchanged (UI testing only)
+#   openai    = DALL-E 3 / gpt-image-1 (~$0.02-0.04/image)
+#   stability = Stability AI SDXL (~$0.01/image)
+#   replicate = Multiple models (~$0.002-0.03/image)
+#   invokeai  = Self-hosted InvokeAI running on another machine
+#   comfyui   = Self-hosted ComfyUI running on another machine
 #
-# RECOMMENDED: Use "replicate" for best quality and model variety
+# GPU QUICK-START:
+#   docker compose -f docker-compose.gpu.yml up --build
+#   (AI_PROVIDER defaults to local_gpu in that compose file)
 # =============================================================================
 
 AI_PROVIDER=replicate
 
+# ── Local GPU settings (only relevant when AI_PROVIDER=local_gpu) ────────────
+# Auto-download HuggingFace models on first request (true/false)
+AUTO_DOWNLOAD_MODELS=true
+# Max diffusion pipelines to keep loaded in GPU memory (each is 2–7 GB)
+LOCAL_GPU_MAX_PIPELINES=2
+# HuggingFace token — only needed for gated/private models
+#HF_TOKEN=hf_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
+# Override auto-selected model for any operation (leave blank = auto by VRAM tier)
+#HF_MODEL_INPAINT=your-org/your-inpaint-model
+#HF_MODEL_TXT2IMG=your-org/your-txt2img-model
+#HF_MODEL_IMG2IMG=your-org/your-img2img-model
+# ─────────────────────────────────────────────────────────────────────────────
+
 # Per-operation provider overrides (optional — blank means use AI_PROVIDER above)
 # Example: use OpenAI for text-to-image (best quality) but InvokeAI for everything else
 #AI_PROVIDER_TXT2IMG=openai