AppLog drops lines written on the way out of the process, so clean shutdown is usually lost #174

AppLog.write hands the line to a background queue (AppLog.swift:28) and returns. When the caller is on its way out of the process, the line races the teardown and is frequently lost.

The site that matters today is CrashReporter.markClean() (CrashReporter.swift:65-68), which runs from the NSApplication.willTerminateNotification observer registered in install() (:45-47). AppKit posts that notification and then exits, so clean shutdown is written only if the log queue happens to get scheduled first.

Measured with an AppKit app reproducing the sequence — enqueue on a serial DispatchQueue, then NSApp.terminate(nil):

runs=100 landed=58 lost=42

Consistent with what is on disk here: the current 1.2M PokeTokenBar.log has 1 launch: line and 0 clean shutdown lines; the rotated 2.0M PokeTokenBar.old.log has 1 and 1.

The consequence is not just a missing line. clean shutdown is the human-readable half of the shutdown record, so the log reads as if sessions end abruptly even when they don't. The marker file itself is removed synchronously (FileManager.removeItem, :66), so the detection in install() :37 is unaffected — this is a diagnostics defect, not a false crash report.

Fix shape: an AppLog.flush() that does queue.sync {}, called on the paths that exit right after logging. Sweep for callers, not just this one — #163 adds a second such site in applicationDidFinishLaunching.

Class, for docs/reference/defect-log.md: an async logger is not a logger on an exit path. Any AppLog.write whose caller reaches exit() in the same turn needs a flush, and the absence of the line cannot be used as evidence that the branch did not run.