Add a notebook that shows how to adjust the max_step when using solvers with time-dependent Hamiltonians.

[hodgestar]

When using time-dependent Hamiltonians where the Hamiltonian is constant for large periods of time, it is possible for the adaption solvers to take very large time-steps that completely step over the small time-dependent part. Periodically our users are very confused by this, so it would be nice to have a notebook that shows the problem and how to fix it.

Related issues:

• sesolve() gives erroneous solutions when given a QobjEvo with an array of coefficients qutip#1403
• Mesolve function returns wrong result using default Ode method qutip#1265
• mesolve results incorrect for certain number of time steps qutip#1945