M8: Chaos Mesh testing suite

[CodeBuildder]

Goal

Deliberately inject failures to test Argus detection and response.

Tasks

• Deploy Chaos Mesh via Helm
• Write chaos experiments:
  • PodChaos: kill random prod pod every 5 minutes
  • NetworkChaos: inject 200ms latency on payment-service
  • IOChaos: inject disk errors on postgres
  • StressChaos: CPU stress on api-gateway
• Verify Argus detects each chaos event
• Build chaos test runner script: chaos/run-tests.sh
• Document expected vs actual detection rate

Acceptance criteria

• All 4 chaos experiments run without cluster damage
• Argus detects at least 3/4 chaos events
• Detection latency < 30 seconds