#12 - Parsing models with Linear Affine Dynamics

[mforets]

Closes #12.
Closes #14.
Closes #17.
Closes #18.
Closes #22.

[mforets]

not urgent, but the tests need to be accomodated to the new convention for the input constraints U and X. 

context: since the type ConstrainedLinearControlContinuousSystem is of the form x' = Ax + Bu, with no constant vector C, we were adding a new column in B with the constants information, and also for the type U putting cartesian product with singleton's whose element takes the value 1.0. this is as workaround that seems to work, but we are not sure it will cause other problems (there can be a dimension mismatch if one location is linear and the other location is affine).

[mforets]

In the last push i changed the design of the inner loops of linearHS, in such a way that they rely on expression-to-lazyset checks and conversion methods.

The drawback of this approach is that it is not optimal performance-wise, since it does the same operation, such as taking a gradient, more than once.

[schillic]

Just return is_linearcombination(lhs) && is_linearcombination(rhs)?

[mforets]

Ready from my side, i've checked that the issues above are fixed, and those which are not, left for later.

This PR adds the new function linearHS that transforms a hybrid system whose fields are expressions, into another hybrid system whose fields are MathematicalSystems types, for the particular case where the dynamics and assignments in each mode are affine.