Extend interpolation functions to integrator/callback

[mjowen]

The current interpolation definition only allows you to interpolate over the final value of a solution. I would like to be able to include an interpolation in a PostprocessCallback so I can see how pressure at a point (for example) evolves over time.
From what I can tell from a glance at the code, it wouldn't be a small change. Since the system data you get during a PostProcessCallback is isolated to a single system, I don't think its possible to do an interpolation.

I could see a simple fix using the custom_quantities interface, which provides u_ode and v_ode, and then if these versions of the interpolation functions were made public, it could be accessed there

From what I can tell, the custom_quantities code is the only public description of u_ode and v_ode. I wasn't sure if you consider using u_ode, v_ode = sol.u[end].x to be valid API (for breaking changes purposes)

[efaulhaber]

I agree, all versions of the interpolation functions with arguments v_ode and u_ode should be public API. They are generally supported and will not be removed/changed soon, so feel free to use them. We only have to add corresponding documentation. I use these functions in the PostprocessCallback every time I want to interpolate something in regular intervals. I just wasn't aware that we never added the documentation.