DrawEvent should be able to manage a given sample

[mbaudin47]

In the PR #39, the DrawEvent class is extended to manage input distributions with dimension greater than 3.

For each cross-cut plot (i,j), the current implementation of the drawSample method uses the marginal bivariate distribution, then generates a sample from this distribution. It would be handy to plot a given sample instead of generating it. Each (i,j) plot would contain the projection of the sample. This would be an extension of drawInputOutputSample().