Rmomal/EMtree: EMtree v2.0

In this second version, EMtree was improved and present with new functionalities:

• Stable handling of big networks (200 to 1000 nodes)
• Set a random initialisation of the EM weights parameters ("random.init" parameter of main function EMtree())
• Run stability selection of probability threshold with the StATS function
• Use a user-provided function for the estimation of the Gaussian covariance matrix in ResampleEMtree ("user_covariance_estimation" parameter)
• Force some unlinked nodes ("unlinked" parameter)

The overall numerical stability was improved thanks to an adaptative conditioning parameter. The vignette Fatala_Net has been updated, and illustrates the new graphic possibilities among which the storing of a layout and the colouring of edges. Finally, the functions "ComparEMtree()" and "compare_graphs()" were deleted.