Variable bounds not updated in branching

[laradicp]

Describe the bug
Branching algorithm adds constraints for variable bounds (in function generate_children), but the bounds of the variable itself are not updated. The variable bounds retrieved from BlockDecomposition.callback_lb(cbdata, var) and BlockDecomposition.callback_ub(cbdata, var) are incorrect.

To Reproduce

1. Define a model with a pricing callback and maxnumnodes = 2
2. Optimize the model
3. Check the branching constraint (variable and rhs)
4. Update lb and ub in the pricing callback for the subproblem in which variable is, using BD.callback_lb(cbdata, variable) and BD.callback_ub(cbdata, variable)
5. Check last values of lb and ub:
   If the branching constraint is greater or equal:

   @test lb == rhs
   

   If the branching constraint is less or equal:

   @test ub == rhs
   

Expected behavior
Test will fail. The bound retrieved is always a perene bound.

Suggestion
setcurlb! and setcurub! in generate_children.