Various possible improvements

[robbiespeed]

When working on adding GC tests #8, I had a few other improvement ideas that came to mind. So I'll suggest them all here, and I'd be happy to make PRs for them at some point if you'd like.

• Isolated benchmarks: One thing I noticed when benchmarking my own experiments is that if there was a memory leak or heavy task scheduled during one benchmark that would affect the results of the next benchmark and the results of other libraries. It would be great if each benchmark case would run in it's own isolated process then return timings to the main process. This is similar to how js-framework-benchmark behaves.

• App benchmarks: If libraries have their own unique API they can be forced into taking non optimal paths in order to conform to the reduced ReactiveFramework API. It would be good to see something like a implementation of js-framework-benchmark without the DOM parts, where libraries have freedom to implement a ReactiveApp API.

• Reporting improvements:

  • Logs/Assertions get collected and reported separately to a log file
  • New column for if there were any assertion failures on a bench case
  • Report crashes on the table as a benchmark failure (Isolated benchmarks will allow this)