Managing clustering effects and learning effects in the design and analysis of randomised surgical trials: a review of existing guidance