pgrep -f wait loops match their own command line

A wait loop like until ! pgrep -f "myscript.sh arg" >/dev/null; do sleep 15; done never exits, because the shell running the loop has that same string in its own argv, so pgrep always finds it. Wait on a marker instead: have the job write a line such as EXIT <code> to a file, and loop on grep -q '^EXIT' file. Or anchor the pattern so it only matches the real process (pgrep -f '^/bin/bash /path/myscript.sh'). Also, when a test harness hooks a JS bridge function (H.toNative = wrap(H.toNative)), code that captured the original function in a closure bypasses the hook. Check the native-side message log instead.