[joeking6989]

What this implements

• A Gradio Space that calls NVIQ's existing deterministic benchmark scorer and evidence-driven audit pipeline (no substitute scores, no simulated provider calls).
• File-based input validation and downloadable JSON/HTML audit artifacts.
• Tests using checked-in NVIQ reference inputs, including rejection of mismatched evidence tracks.
• GitHub Actions deployment to existing Noct-Tech/NVIQ, sourcing UI and engine from one repository with an explicit staged distribution.
• Space metadata and deployment documentation.

Deployment prerequisite

Add fine-grained Hugging Face HF_TOKEN secret in this GitHub repo with write permission to Noct-Tech/NVIQ. The new publish workflow runs on merge to main and after its regression checks; the existing static placeholder files in the Space will be replaced when sync succeeds.

Scope

Only evaluates supplied JSON input, not live provider API calls. Never treat packaged reference fixtures as real provider measurements.

Review checklist

• Existing Python CI passes on this PR
• tests/test_space_service.py passes
• HF_TOKEN secret configured (only before deploy)
• Founder merges PR
• Space reaches Running with the new Gradio interface
• Execute live input validation using authorised non-sensitive uploads