On a possible bias in elemental carbon measurements with the Sunset thermal/optical carbon analyser caused by unstable laser signal