[WIP] Reintroducing cholesky

Created by: tgiani

Hi, here as discussed in the last pc I m trying to do the following:

1. reading the theory cov matrix from file at the level of ComputeCovMat_basic in chisquare.cc, using the function read_total_covmat (which btw should now be called read_theory_covmat)
2. construct the total cov matrix inside ComputeCovMat_basic, adding to the additive errors also the part coming from the theory cov mat, if present
3. performing the replica sampling doing a gaussian sampling using just the additive part of the total cov mat, and adding the mjultiplicative uncertainties by hand, like in the old implementation. The gaussian sampling is performed using cholesky decomposition
4. use the same function ComputeCovMat_basic to build the fit covariance matrix in Experiment::LoadFitCovMat

libnnpdf/src/dataset.cc

@@ -75,7 +75,7 @@ DataSet::~DataSet()
  */
 void DataSet::GenCovMat() const
 {
-  fCovMat = ComputeCovMat(*this, fT0Pred, fWeight);
+  fCovMat = ComputeCovMat(*this, fT0Pred, false, " ", {}, fWeight);
   fSqrtCov = ComputeSqrtMat(fCovMat);
 }