Compute forward map advection after computing velocity to benefit from more optimal timestepping methods

[Arcadia197]

With having the backwards map already avaiable, we could move the forward map advection after the computation of the stream hermite similar to the particle advection. Then we could use heavily simplified schemes to advect everything.

This would then basically mean, that forward map advection and particle advection are kind of similar, with only the forward map benefitting from remapping and being represented on a structured grid

[Arcadia197]

This is very straight forward.
Only thing that needs some adaption is, that currently both maps are remapped before computing the stream function. This has to be changed, so that the remapping still occurs after the computation of the forward map, which should be after computing the stream function.