Grid virtual casing

[missing-user]

Addressing issue #32
I iteratively increase the resolution to get an estimate of how well the integral has converged, and could adaptively refine the grid until a given epsilon is reached. For now it just prints the error estimate and doesn't terminate if gBnstol isn't reached.

For the rotating ellipse test case at a given accuracy, it is approximately 10x faster than the current implementation.

Possible improvements

• Interpolate the precomputed surface currents when performing the integration, to have an "even finer grid" or possibly use an adaptive integration routine ontop of the interpolated values as @lazersos suggested.
• Possibly reuse the computed surface currents across multiple vc iterations, and only recompute the outer integral with the $1/r^2$ term.
• Add test cases comparing the implementations