Remove obico's GPU reservation to free VRAM for ollama

obico's own comment already notes its :cuda tag falls back to CPU
inference on this driver/GPU combo, but it was still reserving 826MB
of VRAM it never used productively. That squeeze was forcing ollama
to offload only 20/29 model layers to GPU, pushing the rest onto CPU
and causing severe latency (multi-minute LLM calls) that stalled
SUB/WAVE's track picking.

Claude-Session: https://claude.ai/code/session_01L7Rwa6guD5wK8F8tWQwcJX

obico/docker-compose.yml

@@ -7,9 +7,10 @@ services:
       - obico-internal
 
   obico:
-    # :cuda tag attempts GPU but falls back to CPU if CUDA version mismatch
-    # (libcudart.so.11.0 required; GTX 1660 SUPER with driver 590 has CUDA 12).
-    # ML inference still works on CPU - adequate for a single printer.
+    # :cuda tag falls back to CPU anyway on this GPU (libcudart.so.11.0
+    # required; GTX 1660 SUPER with driver 590 has CUDA 12) - adequate for a
+    # single printer. No GPU reservation here so it doesn't reserve VRAM
+    # ollama needs for its own model (was forcing partial CPU offload there).
     image: ghcr.io/imagegenius/obico:cuda
     container_name: obico
     restart: unless-stopped
@@ -33,13 +34,6 @@ services:
       - /srv/obico/config/model_cache:/model_cache/ml_api
     ports:
       - "3334:3334"
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              count: 1
-              capabilities: [gpu]
     networks:
       - npm-network
       - obico-internal