Update benchmarks

[ama0]

• Add sc_graph_aligned bench to compare aligned and spread data distribution
• Add ecs_pos_vel_spread to have a more realistic bench for entity composition (not only aligned data)
• Add bench_setup to have a common file with all numbers and stuff

Note: I changed the N_POS and N_POS_VEL values to 15,000 and 5,000 to make the benchmark runtime a little bit higher and better show the difference between spread and aligned data.
With small numbers like 9,000 and 1,000 the difference is smaller then cargo bench's performance fluctuation. Of course the difference becomes more distinct with higher values. It's just a balance between a long wait and more obvious numbers 🙂

[vitvakatu]

👍

[vitvakatu]

👍