feat(main): requestSingleInstanceLock to prevent PTY loss on AppImage replace

[JeanBaptisteRenard]

Problem

On 2026-05-23, replacing the AppImage at ~/Applications/Switchboard.AppImage while Switchboard had active node-pty sessions in the running Electron process killed those sessions. The OS launched the new binary, which initialised a second Electron process. The two instances raced and the running PTYs were orphaned or killed.

Fix

Electron's app.requestSingleInstanceLock() is the standard pattern for this:

• The first instance acquires the lock and continues to run normally (all PTYs stay alive).
• Any subsequent launch (e.g. the new AppImage binary after an in-place replace) fails to acquire the lock, calls app.quit() immediately, and exits without touching any PTY or DB.
• A second-instance listener on the first instance restores and focuses the main window, giving the user visual confirmation the app is still running.

Changes

• main.js (lines 1601–1673): call app.requestSingleInstanceLock() before app.whenReady(); wrap app.whenReady() and all startup code in the else-branch so it only runs for the true first instance; register app.on('second-instance') to restore/focus mainWindow.

Testing

• All 45 unit tests pass (node --test)
• ESLint: 0 errors, 207 warnings (same pre-existing warnings as baseline — none introduced by this change)