Add public validation health dashboard pilot

[Brandon Miller (brandom-msft)]

Summary

• add a two-sample dashboard pilot for the C# and Python quickstart/chat-with-agent samples
• render linked L3/L4 status and last-run columns from normalized public validation results
• keep missing results explicit as Never run
• add credential-free renderer coverage to scripts-selftest

Dashboard contract

The renderer accepts normalized result JSON and maps each completed level to:

• pass → ✅
• sample failure → ❌
• infrastructure/error → ⚠️
• no result → ⚪ Never run

The pilot is intentionally limited to two samples. Both currently have L3 commands and no L4 declarations, so the seeded dashboard shows Never run for all cells until a separate validation producer publishes results.

Scope boundary

This PR owns only the public dashboard presentation and its two-sample pilot configuration. It does not add a validation workflow, schedule, issue publisher, strike counter, or quarantine behavior, and it does not change .github/workflows/validate.yml or the required trusted job.

Validation execution, normalized artifact production, and delivery to the dashboard are being planned and implemented separately under P4.1 so they can consume the P5-owned transition manifest and preserve the frozen cold-L4 and failure/error contracts.

Tracking

ADO 5505679: https://msdata.visualstudio.com/Vienna/_workitems/edit/5505679

Public dashboard: #892