Update comparing_greedy_vs_analytical.cpp

[camel-cdr]

On my system I get the following benchmark results:

$ ./a.out 1337 500 1000000
Testing with 1000000 points, 500 times...
run_number: rejection_2d_ms, analytical_2d_ms, ziggurat_2d_ms, rejection_3d_ms, analytical_3d_ms
1: 36, 37, 17, 55, 101, 
2: 22, 36, 16, 52, 102, 
3: 22, 37, 16, 52, 107, 
...

See my writeup for an explanation. It's probably possible to use a similar approach for sampling from spheres, but I haven't looked into that yet.