Potentially reduce state variables #45
Description
If the wake state variables were to be splined based on cumulative length (so add up the panel lengths along a wake sheet), it might be possible to capture the wake state variables at sufficient accuracy with much fewer state variables in the solve. This could speed up the jacobian-based solvers quite a bit, which could be nice.
I would probably go about this by using the influence_length vector in the wake_vortex_panels object to obtain the full wake sheet length as well as the non-dimensional positions of the wake panel nodes or control points (depending on the residual being used), then I'd spline the strengths along each sheet with respect to distances along the sheet, then sample them at the nodes or control points as needed inside the solve.
In the original setup, I'd probably have an automated spline definition function that defined the spline points along the wake sheets in some intelligent way, perhaps considering the amount of curvature in the sheet and adding more spline points there, or maybe I'd leave it to the user as an addition to the paneling_constants object.