Initial release with the following content:
* Model: MADX via [cpymad](https://pypi.org/project/cpymad/) backend
* Inverse modeling: linear optics from closed orbits (LOCO)
* Parameters:
* Quadrupole and multipole errors
* BPM/steerer gain errors
* Any variables that can be specified in MADX (used via `:=`);
this allows to model, for example, BPM/steerer rolls, longitudinal shifts, etc.
* Jacobian computation:
* Finite-difference approximation
* Analytical formulas
* Optimizers:
* Nonlinear least squares via
[`scipy.optimize.least_squares`](https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.least_squares.html)
* Gauss-Newton
* Feedback iteration
* Allows to connect custom optimizers