Open-door policy versus treatment-as-usual in urban psychiatric inpatient wards: a pragmatic, randomised controlled, non-inferiority trial in Norway