feat: add CLIP embedding service sidecar for GPU machine

FastAPI service that loads ViT-L-14 model and exposes /embed,
/similarity, and /batch-similarity endpoints. Runs on RTX 3090
with NVIDIA GPU reservation. Includes 9 cosine similarity tests.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

clip-service/docker-compose.yml

@@ -0,0 +1,27 @@
+services:
+  clip-service:
+    build: .
+    container_name: clip-service
+    restart: unless-stopped
+    ports:
+      - "5200:5200"
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: 1
+              capabilities: [gpu]
+    mem_limit: 4G
+    memswap_limit: 4G
+    healthcheck:
+      test: ["CMD-SHELL", "curl -sf http://localhost:5200/health || exit 1"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 120s
+    logging:
+      driver: "json-file"
+      options:
+        max-size: "50m"
+        max-file: "3"