feat: Post-Generation Verification — auto-run and heal after test generation (QUA-647)

[Desperado]

Summary

After generate_test_code returns a script_id, the agent previously reported success without running the test. In dogfooding, 4 of 26 generated assertions had signature mismatches or wrong regex behavior that a single run would have caught immediately.

Adds a Post-Generation Verification section to both system prompts (professional and cat mode), instructing the agent to treat generation as a three-step pipeline:

1. Run — execute immediately after generation (run_test for Playwright/Cypress, run_native_test for pytest/rust/go)
2. Heal — if it fails, apply the existing Heal loop (up to 3 fix cycles, already in the prompt)
3. Report — ✅ generated and verified or ⚠ generated — N issue(s) remain after 3 fix attempts

The new section sits directly above the existing "Test Healing" section so the two read as one coherent generate → verify → heal pipeline.

Closes QUA-647

Test plan

• Ask the agent to generate a test — it should immediately run it after generate_test_code returns
• If the generated test fails, it should enter the heal loop and attempt to fix (up to 3 cycles)
• Agent should never report a script_id as "done" without first running it

🤖 Generated with Claude Code

[qualitymaxapp]

✅ QualityMax Pipeline

Gate	Result
🔍 AI Review	✅ Clean
🧪 Repo Tests	✅ 298/298 passed (go)
🤖 AI Tests	✅ 28/32 passed

_{Powered by QualityMax — AI-Powered Test Automation}