fix(desktop): ping only the active connection at startup

[huhamhire]

Problem

At startup, ConnectionRuntimeController.ping() looped over every configured connection to refresh its remote identity (currentUser) and probe serverVersion. For non-active connections this is wasted work:

• Their identity has no UI consumer — app:connections filters the summary to the active connection (services/app.ts), and the connection config schema carries no user field, so the connection list can't render it either.
• The persisted identity of a non-active connection is only used to prewarm an in-memory adapter cache, which is refreshed anyway once that connection becomes active.

Meanwhile the cost is real: extra startup network requests to every connection, and — for an unreachable / misconfigured connection (e.g. a TLS cert altname mismatch) — a adapter ping failed WARN on every launch for a connection that isn't even in use.

Change

Narrow ping() to the active connection only (find(active) + early return). Because the loop now handles a single connection, the pervasive isActive branches collapse (behaviour-equivalent, simpler).

Switching the active connection still re-pings the new active one: the settings-save path sets active_connection_id then calls reconfigureConnections() → reconfigure() → ping(). So on-demand identity refresh on switch is preserved — no separate on-switch hook needed.

The active connection's first-sync trigger and PAT fallback (sync-once-even-if-ping-fails) are unchanged.

Effect

• Startup pings only the active connection; no startup network churn for the rest.
• A broken non-active connection no longer pollutes the startup log with a WARN.
• Active-connection identity refresh / first sync: unchanged.

Verification

npx nx typecheck desktop and npx nx lint desktop both pass.

🤖 Generated with Claude Code