Parser profiler

[RGHenderson]

Closes #1996

[msujew]

Thanks, I'll take closer look at this next week!

[RGHenderson]

Thanks. Happy to move things around and adjust behaviour as required - have set aside some time next week to get this in if possible. Just let me know :)
It's largely just a revival of the PR from march by @ydaveluy at this stage.

[RGHenderson]

A small success story from use of this profiling tool:
We had a collection of languages all using a common expression base language. Some profiling of chevrotain itself led us to believe performance was being impacted by keyword-allowing ID rules. Use of this tool showed us that actually we were spending almost all the time parsing a slightly-incorrectly written expression rule.
From 12s in a stress test down to 25ms!

[msujew]

Thank you, I think this is already looking really nice! I believe there are a few more places where we could benefit from the profiling support (lexer, indexing), but these can be implemented separately.

Please see my comments below.

[RGHenderson]

@msujew thanks! Hopefully those should be addressed.

[msujew]

Thanks, I did some additional manual testing, and this works really well.