Should we standardize the output of our benchmarks?

[Oppen]

I've been toying with the idea of making CRoaring benchmarks print their results in a format compatible with the benchstat tool we use in the Go world for comparing different runs. I think it would help us make more informed decisions when optimizing.

Related to this, after we tackle #284 we should be able to use an allocator that keeps track of how many allocations we make and how much memory we use for a given operation, which could further aid comparison.

[lemire]

Good idea.