incompatibility with nakatamaho/mpack

[dimpase]

It is currently not possible to build sdpa-gmp with mpack from nakatamaho/mpack:

sdpa_linear.cpp: In static member function ‘static mpf_class sdpa::Lal::getMinEigenValue(sdpa::DenseMatrix&, sdpa::Vector&, sdpa::Vector&)’:
sdpa_linear.cpp:155:74: error: invalid conversion from ‘mpackint* {aka long int*}’ to ‘mpackint {aka long int}’ [-fpermissive]
     Rsyev("NonVectors","Lower",N,aMat.de_ele,N, eigenVec.ele,workVec.ele,&LWORK,&info);
                                                                          ^~~~~

indeed, Rsyev has a different signature in nakatamaho/mpack, a correct call would use
take the type of LWORK rather than the type of the pointer to it.

---

another error is probably fixable to tweaking headers:

sdpa_io.cpp: In static member function ‘static void sdpa::IO::printLastInfo(int, sdpa::AverageComplementarity&, sdpa::RatioInitResCurrentRes&, sdpa::SolveInfo&, sdpa::StepLength&, sdpa::DirectionParameter&, sdpa::Residuals&, sdpa::Phase&, sdpa::Solutions&, double, sdpa::InputData&, sdpa::WorkVariables&, sdpa::ComputeTime&, sdpa::Parameter&, FILE*, FILE*, bool)’:
sdpa_io.cpp:1078:23: error: ‘mpf_approx_log10’ was not declared in this scope
   mpf_class digits = -mpf_approx_log10(abs(PDgap/mean));
                       ^~~~~~~~~~~~~~~~