Qgopt

In this new release, we added

1. New manifolds: the manifold of density matrices, the manifold of Choi matrices, the manifold of Hermitian matrices;
2. Documentation;
3. Test of manifolds;
4. Some experimental stuff related to the quantum natural gradient (is not documented yet);
5. Corrections of a lot of small bugs.

- Riemannian Nesterov Momentum optimizer
- Fix small bugs

This release includes two types of manifolds: S++ and Stiefel. For the Stiefel manifold one has the following types of retractions:
- svd based,
- Cayley retraction,

the following types of metrics:
- euclidean,
- canonical,

For the S++ manifold one has two types of metrics:
- log Cholesky,
- logarithmic,

and exact geodesics and parallel transport for them.
This release allows performing optimization using the following list of first-order algorithms on a manifold:

- gradient descent,
- gradient descent with momentum,
- Adam,
- AMSGrad.