Check performance implication of tracking breakpoints

[jeromekelleher]

We currently track all breakpoints during the simulation using an AVL tree. We should profile once #2121 is done and see if this is a significant amount of our time.

If it is, we can make the recombination breakpoint tracking optional and turn it on in the places that we use it in our statistical tests.

@GertjanBisschop would you mind reporting back on what the perf report for a large simulation is wrt to breakpoint tracking?

[jeromekelleher]

Ah, we also use this in mspms so that we have strict compatibility with ms. So, we would need to turn it on for this application.

[jeromekelleher]

Closing this as it's unlikely to get picked up, and we didn't actually make any gains in terms of perf by dropping the per-segment tracking.