feat(fault-injection): Add test helper utilities and package setup

[nv-oviya]

Overview:

Test utility functions for fault injection scripts: colored terminal output, HTTP request helpers, response validation, and environment-based configuration. Works both locally and in-cluster.

Details:

• Add client/test_helpers.py (322 lines): Test utility functions

  • Colors class: ANSI color codes with TTY auto-detection
  • get_config_from_env(): Load API_URL, FRONTEND_URL, APP_NAMESPACE from environment
  • check_frontend_reachable(): HTTP health check with retry
  • send_completion_request(): Send OpenAI-compatible requests with retry logic
  • validate_completion_response(): Validate response structure and content

• Add client/setup.py (35 lines): Python package configuration

  • Dependencies: httpx, kubernetes, pyyaml, requests
  • Enables pip install -e client/

• Add client/conftest.py: Pytest configuration

Where should the reviewer start?

test_helpers.py

• Lines 28-114: Colors class with TTY auto-detection
• Lines 116-180: get_config_from_env() loads configuration from environment
• Lines 182-220: check_frontend_reachable() health check
• Lines 222-280: send_completion_request() inference helper
• Lines 282-322: validate_completion_response() validation

setup.py

• Standard package setup with dependencies

Key features:

• Colors automatically disable for piped output (logs, CI)
• Auto-detects in-cluster vs local execution
• Retry logic for transient failures

Related Issues:

• Relates to: Foundation for fault injection test scripts

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[nnshah1]

I think this one also overlaps with exsiting functionality - we can converge later - but I think if we get the functionality in but leave out the end to end tests -would be a good start