kstate is not needed anymore as all information is found in M_.lead_lag_incidence
See #1653

— Use MX_HAS_INTERLEAVED_COMPLEX for testing for the new API

— Use the typed access functions under the new API, with one exception: we do
  not use mxGetDoubles() instead of mxGetPr(), because it would just complexif
  the code without any value added

For more details, see:
https://fr.mathworks.com/help/matlab/matlab_external/upgrade-mex-files-to-use-interleaved-complex.html

[skip ci]

Ref preprocessor#6

[skip ci]

This is only the interface. The actual implementation and the documentation are
still missing.

Ref #1649

[skip ci]

As a consequence, remove the --enable-openmp option of the configure script.

Testing shows that it is slower than the BLAS path.

It constructs the stacked residuals and jacobian of the perfect foresight
problem.

It is an almost perfect replacement for the perfect_foresight_problem.m
routine, while being much more efficient.

Note however that the DLL never return complex numbers (it instead puts NaNs at
the place where there would have been complex). This may create problems for
some MOD files; the algorithms will need to be adapted to use a more
line-search method.

Do not use varargin.