Clean and skimmed rewrite of the r-adaptation branch

I have re-written the r-adaptation code that was so old it became complicated to update it to master. I took advantage of that to skim the code. There's no only 1 (vs 3 previously) way to run r-adaptation: the user has to provide a list of point with an associated scaling factor. VarOpti then interpolates between these points and applies scaling to each element. I will add an example test for better understanding.