S-rep's DL

[bpaniagua]

Speed up measures for S-Reps: We plan to use DL to accelerate the fitting of S-Reps when large amounts of
data are available. This DL network will learn the best possible fit for a generic S-Rep, to be used as a preprocess
before both the learning stage and for target cases used for testing. Users with GPUs will be able to shorten the
fitting times, which could be important for large sample sizes.

@vicory for Ninad's internship project