Solution data abstraction

[koehlerson]

For time dependent problems and cell data such as fluxes and stresses the solution data needs some abstraction, e.g.

abstract type AbstractSolutionData end
abstract type AbstractStaticData <: AbstractSolutionData end
abstract type AbstractTimeDependentData <: AbstractSolutionData end

...

with a proper interface, such that static and dynamic data as well as point and cell data can be distinguished while dispatching

[termi-official]

One of my thoughts on the design here is that we have some conversion from a transient/dynamic/time-dependent/instationary data source to a static/stationary/steady-state, possibly via temporal interpolation operators, all of which are some type of source. Then we have different classes of filters, which again can be sources, such that we are able to chain filters. This allows us to decouple the data processing portion from the actual visualization.

[termi-official]

Maybe we should opt for a trait system and open dispatches here, so external tools which are built on top of Ferrite can shove their data into the visualization pipeline.