Run sine problem with tourny size 7, 0.1% elitism, and big population (10,240)

[NicMcPhee]

We should re-run the "weak" configuration of the sine problem (tourny size 7 and 0.1% elitism), but with the bigger population (10,240) instead of the smaller (1,024). That would make some of the success plots more comparable since the population size really has a big impact on things like success rates.

It's not clear whether the effects will still be as similar with the larger population, but we should get the data and see.

[NicMcPhee]

Increasing the pop size definitely improves the performance by quite a lot (no surprise). The different bias levels are all essentially the same at the end of the 100 generations, but interestingly there is definitely a spread around 15-20 generations that is a really clean ``more bias is better'' demonstration except for the fact that 0.75 and 1.0 are essentially the same.

Not entirely sure what we want to do with this. Should we be doing all the "weak" configuration runs with pop size 10K to try to level out that part of the playing field?