Benchmark Scripts

[aaravpandya]

Some useful scripts to run benchmarks. Supports the following benchmarks

1. Normal - Runs and calculates AFPS for gradually increasing number of worlds.
2. Random - Runs random trials for each number of world to get a smoother curve.
3. Binned Datasets - Bins the dataset by number of agents and then runs trials for each bin. Generates an AFPS vs Num agents graph.

All results are saved to a CSV.

[SamanKazemkhani]

@aaravpandya Can you rebase this over main? I'd like to use it to benchmark #106.