Observational and genetic evidence disagree on the association between loneliness and risk of multiple diseases