v3.9.1 — the dashboard answers before the diagnostics do

The dashboard answers before the diagnostics do

Fixed

• The dashboard's Brains table and stats cards no longer block on full-brain
  diagnostics. With the server running away from its database (a k3s pod
  talking to SurrealDB over WebSocket), a cold
  DiagnosticsEngine.analyze costs ~100 s and a warm one ~10 s — the browser
  gives up long before either finishes, and the overview rendered "No brains
  found" over zero counters although the data was fine. /api/dashboard/brains
  now returns as soon as the per-brain counts are known (grade/purity render
  at their defaults until computed), a previously computed table is served
  instantly while a single background pass refreshes it, and an expired
  diagnostics report is served stale and recomputed once in the background.
  The cache TTL now paces the refresh instead of gating the caller.

Full Changelog: v3.9.0...v3.9.1