fix(doctor): restore compact eleven-row System Doctor

[levineam]

Summary

System Doctor now reads as one compact status board: every selected subsystem has one green check, red X, or warning, while degraded rows explain the problem and the next useful action. This replaces the repeated PASS/FAIL transcript and restores the eleven-row Memory Stack Doctor roster, including separate GBrain core and semantic-coverage rows.

This corrects the presentation introduced by #277. Private snapshot generation and Telegram delivery remain in SUP-3899; the separate Desktop rendering remains in SUP-3900.

Before and after

Before:

Core:
❌ FAIL workspace-files — repair needed

Selected optional components:
⚠️ WARN SearXNG — warning
✅ PASS GBrain — healthy

NOT READY — repair needed

After:

Core
❌ workspace-files — Required workspace file is missing. Fix it, then rerun Doctor.

Services
⚠️ SearXNG — No search results. Run a real search, then rerun Doctor.

Memory
✅ GBrain core
✅ GBrain semantic coverage
✅ Lossless Claw
✅ QMD search
✅ memory-wiki
✅ Notes & provenance
✅ Recall evaluation
✅ Scheduled maintenance
✅ Reviewed runtime
✅ Automatic repair
✅ Telegram follow-up & proof

Contract boundary

The structured jarvos-system-doctor-report/v1 receipt and its reason codes remain intact. The exact eleven-row producer roster is now jarvos-system-doctor-facts/v2; obsolete ten-row v1 snapshots fail closed. A complete v2 roster is authoritative, so the legacy Memory aggregate is not rendered or counted twice.

Validation

• Complete npm test suite passed on commit 71d79fd.
• Focused Doctor module, CLI smoke, and runtime-neutral tests passed.
• Independent correctness, testing, API-contract, reliability, and adversarial reviews were resolved; no public-code findings remain.
• This PR does not prove a regenerated private v2 snapshot, Telegram parity, scheduler activation, or Desktop acceptance.

Related: SUP-3896