Use Julia package for benchmarking

[Jingru923]

In issue #462, regression test is set up for comparing output of a run with the benchmark.

@visr suggests to add runtime into benchmark and make use one of existing Julia package to illustrate the benchmark result better

[Huite]

This one? https://github.com/JuliaCI/PkgBenchmark.jl

I'm aware of asv in Python, and it seems there's quite a lot involved into getting consistent and commensurate benchmarks.

[visr]

PkgBenchmark.jl has been around for a while but doesn't seem very actively maintained. Also for CI integration it recommends an unmaintained package.

PkgJogger.jl seems like a better maintained alternative that has nicer local workflows, with easier CI integration.

Though less mature, Lilith's work in https://chairmarks.lilithhafner.com/v1.2.2/regressions seems very nice as well. Perhaps I'd try that first.

In terms of published benchmark tracking, there is https://github.com/SciML/SciMLBenchmarks.jl but comparing commits over time like https://lux.csail.mit.edu/benchmarks/ also looks nice.

EDIT: there is now also AirspeedVelocity.jl: https://discourse.julialang.org/t/easy-github-benchmarking-with-new-airspeedvelocity-jl/129327