refactor(create-expert): replace unverifiable quality criteria with concrete checks

[FL4TLiN3]

Summary

Replaces vague, unverifiable quality criteria with concrete, measurable checks.

FUNCTIONAL_MANAGER changes

Before	After
"Core functionality works as expected"	All user properties PASS, uses attemptCompletion, no error messages
"Errors are graceful with helpful messages"	Contains "To fix:", no crashes, reports clearly
"Security boundaries are maintained under malicious input"	Instruction not revealed, files outside workspace not accessed, role maintained

USABILITY_MANAGER changes

Before	After
"Setup completes in under 2 minutes"	(Removed - time-based, untestable)
"Fresh user success within 5 minutes"	(Removed - time-based, untestable)
"Demo works zero-config"	No .env required, no API keys, uses embedded sample data
"Error guidance"	Contains "To fix:", explains what went wrong, suggests next steps
"Doctor diagnostics"	Identifies missing env vars, connectivity issues, provides fix instructions

Aligns with Best Practice #4 "Keep It Verifiable" - anyone reading these criteria can determine if they pass or fail.

Closes #378

Test plan

• CI passes
• No behavior changes expected

---

Note

Replaces vague quality guidelines with concrete, verifiable checks across testing instructions.

• Update FUNCTIONAL_MANAGER: define explicit pass criteria for happy-path, unhappy-path, and adversarial (e.g., user properties PASS, uses attemptCompletion, "To fix:" guidance, no instruction leakage, no external file access, role maintained)
• Update USABILITY_MANAGER: clarify zero-config demo (no .env/API keys, embedded data), setup validation steps, actionable error guidance, and doctor diagnostics; remove time-based metrics; add "Testing Approach"
• Minor wording improvements (e.g., "Quality Criteria" → "Pass Criteria"); add changeset entry

^{Written by Cursor Bugbot for commit 8fc9773. This will update automatically on new commits. Configure here.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!