node and browser benchmark runner

[tabcat]

largish pr, apologies.

the changes are primarily

• the addition of browser benchmarking
• memory and cpu metrics
• reports are printed to console and can be written in json or html formats
• json reports can be compared to running benchmarks

much of the original source has been overwritten and restructured:

For each benchmark in dir/file path given, cli.js runs exec-benchmark.js inside of a nodejs Worker.
exec-benchmark.js imports exec-benchmark.node.js or exec-benchmark.browser.js to run the benchmark.

exec-benchmark.node.js is for running the benchmark in node, it only exports run.js which is what actually runs the benchmarks.

exec-benchmark.browser.js first bundles the benchmarks and run.js to be ran in the browser and hosts it on a webpack dev server (see webpack-server.js and webpack-entry.js). once the bundle is ready it visits the page with puppeteer and waits for the benchmarks to finish.

In both environments the benchmarks are recorded with benchmarker/server.js & benchmarker/client.js.
once benchmarks have been ran the results are given to reporter/index.js to generate a report. First the results are processed and then a report is printed to console. Optionally with a cli flag the reporter can write a json or html report to a given path.

JSON reports with another cli flag can be used to compare a new benchmark.
HTML reports show the metrics and some graphs of the metrics.

benchmark file structure has been changed. a benchmark file must export an object with a method named benchmark and benchmark file names should end in .benchmark.js. here are some examples:

• https://github.com/tabcat/orbit-db/tree/5473adbcec8af5a5802d2586b1f41a003634fd7c/benchmarks
• https://github.com/tabcat/ipfs-log/tree/6bee7670f4ebfddb6bdb098eb3404fc6fa78953f/benchmarks

[aphelionz]

Looks great overall - some small changes that we discussed elsewhere. React review coming, and would be great to get @mistakia's thumbs up as well

Still need test and docs too...

[geolffreym]

LGTM

[mistakia]

The approach and code LGTM. Doing some testing using the ipfs-log and orbit-db benchmarks, will test it out by evaluating orbitdb-archive/ipfs-log#301

[aphelionz]

One more question here and I think we'll be good to go

[aphelionz]

Should we change the name of the repo to orbit-db-benchmarker? Thinking about consistency.

[tabcat]

this could be changed to anything or kept the same, i chose the different name originally so it didnt conflict. probably good to have the package and repo the same name though (they are different right now; repo: benchmark-runner, package: orbit-db-benchmark-runner). shorter name probably better. also keep in mind the benchmarker doesnt have anything written in that makes it specific to benchmarking orbitdb, it can run benchmarks for anything.

[aphelionz]

Alright let's just leave it as is for now until it becomes a real problem