Classification without fix_intercept now included in all files

no longer possible with libsvm

don't reset inds, just don't loop in initial inds.back() < 0

The new iterate function can lead to a -1 for the initialization of inds when n_sis == 1 and
n_ranks > 1. This checks this and ensures it can run with valid inds

forgot to change back

add a copy constructor to make it work, and remove sharkML from cmake

Something is wrong with sharkML in some enviroments

make it more efficent

Add Support for Parallel Build When Building Boost

Closes #1

See merge request tpurcell90/cpp_sisso!11

With these changes, building with one processes takes 8m37.474s on my
laptop, and building with four processes takes 3m16.221s.  Note that the
downloading and unpacking of Boost takes a significant amount of time,
and this step is non-parallelizable.

boost mpi does not seem to like impi 2019

forgot a pair of parens when needed

some overridden functions were not returning when they should be

Create new functions that removes copy/paste from the python constructors

1) Width reduced to 1e-5
2) SVM done with a C=100
3) SVM done with Kernel SVM but with a linear Kernel
4) Fixes to how algorithm breaks ties
5) Added a python SVM function
6) Modified Model python interface to accomadate SVM functions

See if this passes now

Projection was only using last residual
Model did not print proper SVM misclassified

don't use svn, and use the same LAPACK/BLAS as the SISSO Code

Maybe also the other issue on the gitlab server

Debug tomorrow

performs better than Fortran in terms of accuracy
SVM is a challenge in terms of timing

use a linear prgramming method to check the feasabilty of a convex
combination of the classes to get an estimate of the number of points
in overlapping convex hull.

Throw error if no new features were generated during a rung

the max time is not actually limiting time, must rethink strategy

Adjsut how Boost/Shark interact with each other
Explicity set library directory to lib not lib64

Compiles but not tested