feat: Phase 2.6 edges migration to Qdrant (TKG-only architecture)

Phase 2.6.1: co_occurrence_edges migration
- build_co_occurrence_edges_from_qdrant()
- Qdrant embeddings → frame grouping → YOLO objects
- Result: 6679 edges (vs 6701 PostgreSQL)

Phase 2.6.2: face_face_edges migration
- build_face_face_edges_from_qdrant()
- Qdrant embeddings → frame grouping → face pairs
- mutual_gaze detection preserved
- Result: 6 edges (exact match)

Phase 2.6.3: speaker_face_edges migration
- build_speaker_face_edges_from_qdrant()
- Qdrant embeddings → trace_id frame ranges
- SPEAKS_AS edge creation

Architecture:
- All edges use Qdrant payload (no face_detections queries)
- PostgreSQL fallback for empty Qdrant
- Estimated 3.6x performance improvement

Testing:
- Playground (3003): ✓ All Phase 2.6 logs verified
- Edge counts: ✓ Close match with PostgreSQL
- Fallback: ✓ Working

Docs:
- docs_v1.0/DESIGN/TKG_PHASE2_6_EDGES_MIGRATION.md
- docs_v1.0/M4_workspace/2026-06-21_phase2_6_test.md

v1.1/scripts/start_qwen3vl_v1.11.sh

@@ -0,0 +1,35 @@
+#!/bin/bash
+# Start Qwen3-VL server
+
+MODEL_PATH="/Users/accusys/models/Qwen3VL-8B-Instruct-Q8_0.gguf"
+MMPROJ_PATH="/Users/accusys/models/mmproj-Qwen3VL-8B-Instruct-F16.gguf"
+LOG_FILE="/Users/accusys/momentry_core/logs/qwen3vl_8086.log"
+PID_FILE="/tmp/qwen3vl.pid"
+
+# Kill existing process if running
+if [ -f "$PID_FILE" ]; then
+    OLD_PID=$(cat "$PID_FILE")
+    if ps -p "$OLD_PID" > /dev/null 2>&1; then
+        kill "$OLD_PID"
+        sleep 2
+    fi
+    rm "$PID_FILE"
+fi
+
+# Start server
+nohup /opt/homebrew/bin/llama-server \
+    --model "$MODEL_PATH" \
+    --mmproj "$MMPROJ_PATH" \
+    --host 127.0.0.1 \
+    --port 8086 \
+    --ctx-size 8192 \
+    --n-gpu-layers 99 \
+    --threads 8 \
+    --batch-size 512 \
+    --media-path /Users/accusys/momentry/output_dev \
+    > "$LOG_FILE" 2>&1 &
+
+echo $! > "$PID_FILE"
+echo "Qwen3-VL started with PID $(cat $PID_FILE)"
+echo "Log file: $LOG_FILE"
+echo "Health check: http://localhost:8086/health"