Support for correlated multi-variate distributions

[marinaevers]

We need a class CorrelatedDistributions that stores a set of distributions or time series and the corresponding pairwise correlation matrices.
This class should store the individual distributions as well the covariances as members.
The functionalities should be as follows:
mean(dim_i) returns the mean of the i-th distribution
cov(dim_i, dim_j) returns the cross-covariance between the i-th and j-th distribution if i not equal to j and the covariance matrix of the i-th distribution if i=j
The sample function should sample from the joint distribution.
If something else is needed, especially for the application to UA-STL, this can of course be discussed.