CPU-parallelization

[Smintheus98]

New feature to accelerate the resampling with CPU based parallelization.

The parallelization is realized by a call to parallel::mclapply instead of the previous lapply.
To get safe parallel random numbers, the RNG is replaced by the L'Ecuyer-CMRG generator, while the initial RNG is stored and reset after the resampling.

This feature breakes the interface of the GOF_model_test class constructor, which now requires two additional parameters n_cores and seed, which are required to set the degree of parallelization and seed the internally setup RNG.
These parameters are also new to the GOF_model main function (which are simply passed to the test class), using default values however.

There is a new section in the README explaining a bit about the parallelization with a short example.