better handling of PCA by allowing giving a single matrix instead of a list with one element.
refactor scaling to simplify the function and adequately handle NA values in scale_lambda1 and scale_inertia.
change AVE computation to match PCA's AVE. The AVE of a component is the variance of the component over the variance of the associated block. Non-orthogonal components are orthogonalized using QR decomposition.
This PR leads to a few improvements:
scale_lambda1andscale_inertia.