Data supporting the article 'Functional diversity metrics can perform well with highly incomplete datasets'