dt in post_explicit! and post_implicit!

[Sbozzolo]

(This is not really an issue, more of a consideration.)

The post_{implicit, explicit}! functions are called with signature: (u, p, t). Therefore, if the post_ function wants to use the value of the timestep dt, that has to be included in p. Now, in my opinion, this is a bit dissatisfying because I think that the integrator should own dt and be responsible for it. Hard-coding dt in p introduces a potential mode of failure in which the dt in the integrator and the dt in p disagree on the value of the timestep.