What can we learn about orographic drag parametrisation from high‐resolution models? A case study over the Rocky Mountains