Lipid monolayers: why use half a membrane to characterize protein-membrane interactions?