Benchmarks

Having a set of standard benchmarks will help us improve performance in systematic way and avoid performance regressions.

It's also going to help making sure Go version is good enough performance-wise.

Some benchmarks proposed by @sparkprime :
dcunnin@casterly:~$ time jsonnet -e 'std.length(["XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" for x in std.range(1,10000)])' > /dev/null

real 0m0.539s
user 0m0.472s
sys 0m0.064s
dcunnin@casterly:~$ time jsonnet -e 'std.length(std.join("\n", ["XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" for x in std.range(1,10000)]))' > /dev/null

real 0m3.675s
user 0m2.928s
sys 0m0.744s

One question is how to present the results (i.e. running time) in a way that can be easily compared across versions/implementations.