Extend rounding options and support sparse computations

[TolisChal]

This PR implements an additional rounding method for convex polytopes based on the log-barrier minimizer (analytic center) and adds support for sparse computations across multiple rounding methods, maximum ball computation, and feasible point computation for H-polytopes. The implementation uses compile-time decisions through if constexpr (requiring C++17) to handle both dense and sparse matrix operations efficiently.

Key Changes Made:

1. New Rounding Method: Introduces analytic center-based rounding as an alternative to maximum ellipsoid rounding
   Sparse Matrix Support: Adds comprehensive sparse matrix support for computational operations using Eigen's sparse matrix types
2. Template Metaprogramming: Uses if constexpr for cleaner compile-time type selection
3. New Computational Framework: Creates mat_computational_operators.h to abstract matrix operations for both dense and sparse matrices
4. Enhanced Testing: Adds comprehensive tests for sparse computations and new rounding methods
   C++17 Upgrade: Updates build system to require C++17 standard