Comments on the justification of the independence assumption in 'Does cycle commuting reduce the risk of mental ill-health? An instrumental variable analysis using distance to nearest cycle path'.