Fix CI test failures by properly configuring data pipeline

[nee1k]

Summary

Fixes the CI test failures where tests expected nodes in Neo4j but found 0.

Root Cause

1. Oracle daemon was connecting to external Kafka broker (cknbroker.pods.icicleai.tapis.io:443) instead of the local CI broker (broker:29092)
2. Kafka security protocol was hardcoded to SSL, but CI broker uses PLAINTEXT
3. CI wasn't waiting for Kafka Connect connectors to be ready before running tests
4. Insufficient wait time for data to flow through the pipeline (daemon → Kafka → Neo4j)

Changes

• oracle_daemon.py: Add CKN_KAFKA_SECURITY_PROTOCOL env var (defaults to SSL for backward compatibility)
• docker-compose.ci.yml: New CI-specific override file that configures:
  • Local Kafka broker (broker:29092)
  • PLAINTEXT security protocol
  • Connection to ckn-network
• ci.yml:
  • Add step to wait for Kafka Connect and Neo4j sink connector
  • Use docker-compose override for CI environment
  • Increase wait time for data ingestion

Test plan

• CI passes "Build CKN" step
• CI passes "Test CKN Oracle Daemon plugin WITHOUT power_monitoring" step
• All 6 node count tests pass (Experiment, EdgeDevice, Deployment, User, Model, RawImage)