Add a direct n-body solver

[namurphy]

An idea that came up while working on the Coulomb collision notebook in #1569 was that it would be useful to have a direct n-body solver. Having such a solver would let us launch a charged particle towards another at some impact parameter and then calculate and visualize the trajectories, which would be a really helpful thing to do when teaching Coulomb collisions. My guess is that we'd be able to implement a direct n-body solver in $∼100$ lines of code (plus tests & docstrings) so the maintenance burden would probably be low.

The solver would need to be compatible with astropy.units, though it should work with NumPy arrays under the hood for performance reasons, and so that we could make better use of Numba. It'd also need to be compatible with plasmapy.particles. The governing forces could be either the Coulomb force or gravitation.

We should also consider alternatives to writing our own n-body solver, such as using one that already exists elsewhere in the pythoniverse and writing a wrapper to make it compatible with astropy.units and plasmapy.particles if it isn't already. We could use then use that solver in the Coulomb collision notebook.

[namurphy]

There are also more complicated algorithms that would offer better scaling, but with the tradeoff of increased code complexity. The scaling of a direct n-body simulator is $𝒪(n^2)$, which would be enough for educational use cases. One example would be the Barnes-Hut method which is $𝒪(n~\log~n)$. The increased code complexity makes me wonder if that would be more appropriate in its own package, which we could import and then make use of. In any case, that's beyond the scope of this issue.