Fix power monitoring test shutdown signal mismatch

[nee1k]

Summary

Fixes the power monitoring test failure where total_cpu_power_consumption and total_gpu_power_consumption properties were not found in the Deployment node.

Root Cause

The EXPERIMENT_END_SIGNAL environment variable was set to test-end-signal, but the sample data (image_mapping_final.json) uses the UUID 6e153711-9823-4ee6-b608-58e2e801db51 as the shutdown signal.

This caused the daemon to:

1. Never detect the shutdown signal in the event data
2. Never call process_summary_events() to send power data to Kafka
3. Result in no power data being written to Neo4j

Changes

• docker-compose.yml: Update EXPERIMENT_END_SIGNAL to match the UUID in sample data
• ci.yml: Increase wait time for power data pipeline (daemon shutdown → Kafka → Neo4j)

Test plan

• Power monitoring test passes: test_power_monitoring_true
• Deployment node contains total_cpu_power_consumption and total_gpu_power_consumption