idynfor first version

According to #1, a minimum viable product would be to:
> first test the interface to have the 3 advantages above and use only Eigen types for the time being. Then we can check the consistency between iDynTree and this interface, then we can move to Casadi.
Another task would be to sort out the semantics differences between pinocchio and iDynTree. (robotology/idyntree#674)
Another macrotask would be to understand how to translate the iDyntree model to the Pinocchio model.
Another task is to create a skeleton, with the CMake machinery which is not trivial

Hence, the subtasks are:
- [ ] [Create a templated interface that mimicks KinDynComputations](https://github.com/dic-iit/idynfor/issues/2)
- [ ] [Understand the differences between ``iDynTree::Model`` and ``pinocchio`` models and allow switching from the former to the latter](https://github.com/dic-iit/idynfor/issues/3)
- [x] [Create a CMake skeleton](https://github.com/dic-iit/idynfor/issues/4)