Extend multiple_shoot loss to multidimensional NeuralODEs

[SCiarella]

Checklist

• [ YES ] Appropriate tests were added
• [ YES ] Any code changes were done in a way that does not break public API
• [ YES ] All documentation related to code changes were updated
• [ YES ] The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• [ YES ] Any new documentation only uses public API

Additional context

This PR extend multiple_shoot to work for NeuralODEs in $d>1$ dimensions, so u and the right-hand-side can be defined as multidimensional tensors.
After implementing the small change in src/multiple_shooting.jl I have also refactored test/multiple_shoot_tests.jl in order to reuse the preexisting test for the new multidimensional case.