How do different ways of using and organising in-house scientific expertise affect the quality and use of scientific evidence and advice in policy-making?