* Introduced the control parametrization notion and three polynomial parametrization (PolyZero, PolyOne, PolyTwoRK)
* Improved the documentation of the actuation model and especially the floating-base actuation
* Improved the efficiency of the RK integrator (added to the benchmarks)
* Improved the documentation of Euler and RK integrators
* Added unittests for checking the analytical derivatives of the contact forces/impulses
* Computed the dynamic feasibility in solvers (also print this relevant information)
* Improved the documentation of solvers
* Removed dynamic memory allocation in ContactModel3D
* Added the notion of terminal (calc, calcDiff) computations (fixed some tiny inaccuracies in the problem formulation + reduced useless computations)
* Added a class to easily profile the computation cost of any block code (included relevant blocks in the solvers and shooting problem)
* Fixed error in the documentation of ActivationModel2NormBarrier
* Added the 2d contact
* Added method to resized solver data (e.g., it is not needed to allocate the biggest control dimension)
* Fixed issue in the Gepetto viewer display
* Improved the CI and fixed a few errors that appears in unittests code compiled with clang
* Removed dynamics memory allocation in LQR action and CostModelResidual
* Removed Travis buildfarm and substituted by ROS one
* Used std::set for contact/impulse active/inactive set (added bindings)
* Added Python bindings to be able to set state dimensions from a Python derived class.
* Added Python bindings of StateNumDiff class