Add post: Evaluating Copilot Studio Agents (Scope, Target, Deliver)

[KarimaKT]

Summary

• New post: Evaluating Copilot Studio Agents: Scope the Value, Target the Value, Deliver the Value
• Walks the Copilot Studio evaluation and analytics suite end-to-end through three lifecycle stages, anchored by the data type at each stage (Starter Data → Generated Data → Real Collected Data)
• Includes practical Mermaid diagrams (the data-flow loop, working/blind set split, change types from the maker, change types outside the maker, recommended grader pairings by bucket) and Eurozone-agent worked example with screenshots
• Adds a "Features worth knowing about" tail section covering DLP setup, grader stacking, single-turn coverage, and automation links to the Quality Gates and Closing the Loop posts

Checklist

• Local server renders correctly (./tools/run.sh)
• All images have alt text and italic captions
• Internal links to related posts: [Quality Gates for Copilot Studio]({% post_url 2026-04-19-copilot-studio-eval-gate-azure-devops %}) and [Closing the Loop]({% post_url 2026-03-29-agentic-improvement-loop %})
• Tags chosen for Chirpy "Further Reading" overlap (evaluation, analytics, test-sets, graders, custom-metrics, agent-lifecycle, roi, regression-testing)
• Header image, no_bg
• No em-dashes; US English

Closes the runway for the end-to-end evaluations post draft.

[KarimaKT]

Superseded by #331 (renamed slug + reworked draft). Closing this one to keep a single PR for the post.