Sometime particles are less than 2R apart without a collision!

[weka511]

I have added an assertion to Collision.get_collisions() to check that we this condition was never violated, and it is. Further investigation has shown that this is just after a collision, so an error in calculation (possibly round-off?)

[weka511]

• I could exclude the events for the two that just collided from calculation of new set of events; they will be put back after they collide with walls. This may not work for torus however.
• Or I could force centres apart when collision occurs, say to R + np.finfo(float).eps.

[weka511]

Forcing the centres apart seems to have eased problem somewhat. Maybe investigate hitting walls?