feat(machine): connectivity auto-recovery + on-screen Retry

A connectivity-type init failure (e.g. "No connected relays" when the box
boots before the network) landed on "ATM Unavailable" permanently: init is
one-shot and the nostr reconnect only helps after a first successful
connect, so a machine never self-healed when internet returned.

Recover by reloading the renderer, which re-runs init from a clean JS
context (no leaked actors/subscriptions) while the main process keeps HAL:
- main.ts: new `app:recover` IPC → reloadRenderer() (resets secretsConsumed).
- hal:init is now idempotent (reuse the existing instance) so the reload —
  and the pre-existing watchdog crash-reload — can't double-open serial ports.
- App.vue: when initError is a connectivity type (not the operator/
  self-clearing states unpaired/awaiting-fees/maintenance), watch for the
  `online` event (recover immediately) plus a 45s backoff safety net, and
  render a kiosk-sized Retry button for a person at the machine.

Preserves pairing + /var/lib state (renderer reload, not a process restart).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

apps/machine/electron/main.ts

@@ -404,6 +404,17 @@ ipcMain.handle('app:relaunch', (): void => {
   app.exit(0)
 })
 
+// Connectivity recovery: reload the renderer to re-run init from a clean slate
+// (fresh JS context → no leaked actors/subscriptions), while preserving HAL in
+// this main process (reloadRenderer resets secretsConsumed so get-atm-secrets
+// works again, and hal:init is idempotent). The renderer calls this when it's
+// stuck on a connectivity-type "ATM Unavailable" and the network returns, or
+// when the operator taps the on-screen Retry (ADR-002 amendment 2026-08-04).
+ipcMain.handle('app:recover', (): void => {
+  console.log('[Recovery] Reloading renderer to re-attempt initialization')
+  reloadRenderer()
+})
+
 // State persistence IPC handlers
 ipcMain.handle('state:load-cassettes', () => loadCassettes())
 ipcMain.handle('state:set-cassettes', (_event, cassettes) => setCassettes(cassettes))
@@ -480,6 +491,17 @@ let pendingBillDenomination: number | null = null
 
 ipcMain.handle('hal:init', async (_event, config) => {
   try {
+    // Idempotent: HAL lives in this (long-lived) main process, but the renderer
+    // re-runs full init on every reload — the watchdog's crash-recovery reload
+    // and the connectivity-recovery reload (app:recover) both re-invoke this.
+    // initializeHal opens serial ports without closing prior handles, so
+    // re-entering it would double-open the validator/dispenser. Reuse the
+    // existing instance instead; its validator event wiring already targets the
+    // (reloaded) mainWindow, so the reloaded renderer keeps receiving bill events.
+    if (halInstance) {
+      console.log('[Electron] HAL already initialized — reusing existing instance')
+      return { success: true }
+    }
     // Override cassette config with DB values (operator may have changed them via atm-tui
     // or via an operator-config publish from satmachineadmin). Pass per-position so the
     // HAL knows about every bay including duplicates of the same denomination — real

apps/machine/electron/preload.ts

@@ -124,6 +124,8 @@ contextBridge.exposeInMainWorld('electronAPI', {
   // then relaunch so the normal boot flow pairs it.
   saveSpireSeed: (seed: string): Promise<void> => ipcRenderer.invoke('state:save-spire-seed', seed),
   relaunchApp: (): Promise<void> => ipcRenderer.invoke('app:relaunch'),
+  // Reload the renderer to re-attempt initialization (connectivity recovery).
+  recoverApp: (): Promise<void> => ipcRenderer.invoke('app:recover'),
 
   applyOperatorCassettesConfig: (
     payload: {
@@ -243,6 +245,7 @@ declare global {
       resetForRepair: () => Promise<void>
       saveSpireSeed: (seed: string) => Promise<void>
       relaunchApp: () => Promise<void>
+      recoverApp: () => Promise<void>
       applyOperatorCassettesConfig: (
         payload: { positions: Record<string, { denomination: number; count: number }> },
         eventCreatedAt: number