Randomized string concat - #216

Thnx @JuanVqz, no worries for the delay :). I've commited suggested!

@mateusdeap could you please have a look to merge it?

@alekseyl @JuanVqz I noticed that this PR uses bmbm and we want all benchmarks to use ips
Could you please tweak this benchmark to use "benchmark/ips"?
It shouldn't affect the results.
Thanks!

@etagwerker Hi Ernesto!

Sorry for the delayed answer. The problem is it will not look as good as we migh expect, because of a randomization )). It will show a much bigger variance (20-30-40% I can't recall properly) over relatively "small" difference.

I might give it a try one more time, but can't promise that it will be soon enough :(. I need an inspiration to dive in to that river again ))).

Can we also add:

[foo, bar].join

It would also make sense to check if there will be different results with more than 2 strings to concatenate. Also, would be nice to see performance comparison of cases like:

"#{foo}-#{bar}"