[WIP] Upgrade from 4 to 8 ingredients

[cvanaret]

The goal of this PR is to upgrade from 4 to 8 ingredients:

• reformulation layer:
  • constraint relaxation strategy
  • inequality handling method
• subproblem layer:
  • local model
  • hessian model
  • regularization strategy
• subproblem solver layer
  • subproblem solver
• globalization layer:
  • globalization strategy
  • globalization mechanism.

Steps:

• [x] introduce a LagrangeNewtonSubproblem ingredient
• [x] introduce InequalityHandlingMethod ingredient
• [x] introduce HessianModel ingredient
• [x] introduce RegularizationStrategy ingredient
• [x] pass LagrangeNewtonSubproblem to QP/LP subproblem solvers
• [x] pass LagrangeNewtonSubproblem to linear solvers
• [x] defer the Hessian computation of the Model to the HessianModel
• [ ] pass LagrangeNewtonSubproblem to globalization strategy
• [x] rename QPSolver and SymmetricLinearSolver into InequalityQPSolver and EqualityQPSolver
• [x] merge classes QPMethod and LPMethod into class InequalityConstrainedMethod
• [ ] handle InequalityQPSolver and LPSolver in InequalityConstrainedMethod
• [ ] fix computation of least-square multipliers
• [ ] allocate and store the linear solver required in PrimalRegularization and PrimalDualRegularization
• [x] allocate memory for matrices only in subproblem solvers
• [ ] create matrix views in subproblem solvers for evaluation of Hessian and Jacobian
• [x] compute number of Hessian nonzeros for memory allocation
• [ ] restore unused ScaledModel