[Optim] Some persistence computations are **slow**

[DavidLapous]

For ex: Matrix vs GudhiCohomology has a 15x performance diff in this notebook.
And the optimization ignoring inf values isn't compatible with GudhiCohomology, and to some extent with Matrix (e.g., with hilbertfunction) cf

• multipers/multipers/gudhi/truc.h

  Lines 344 to 352 in ca2417c

  	if constexpr (!PersBackend::is_vine && ignore_inf) {
  	for (auto &i : out_gen_order)
  	if (one_filtration[i] == MultiFiltration::Generator::T_inf) {
  	// TODO : later
  	// int d = structure.dimension(i);
  	// d = d == 0 ? 1 : 0;
  	// if (degrees.size()>d || degrees[d] || degrees[d-1])
  	// continue;
  	i = std::numeric_limits<typename std::remove_reference_t<decltype(out_gen_order)>::value_type>::max();

• multipers/multipers/multi_parameter_rank_invariant/hilbert_function.h

  Line 508 in ca2417c

  slicer.template compute_persistence<false>(); // BUG : cannot put this to

• multipers/multipers/multi_parameter_rank_invariant/rank_invariant.h

  Line 231 in ca2417c

  slicer.template compute_persistence<false>(); // PUT THAT TO TRUE...

• multipers/multipers/gudhi/mma_interface_matrix.h

  Lines 187 to 191 in ca2417c

  	for (auto i : *permutation_) {
  	if (i >= boundaries.size()) {
  	c++;
  	continue;
  	}

The inf trick with simplextree-only works with GUDHI/gudhi-devel#896.

@hschreiber I let you take care of that 😉