Note that the issue that was shipped with M58 has since been fixed.  However, it's still true that a navigation on one Tab will cause the entire session history of every other Tab to be re-serialized along with it.  Performance issues have improved, but are not fully optimized quite yet.

Code has been serialising all WebState at each change for a really long time (issue 661986 was create from a TODO by original code introduced by pinkerton when the serialisation on iOS was introduced). I don't think thinking this bug should block developping perf test, in fact I think it should be the other way around, we should wait for making changes that could impact perf before we can measure them.

Hi Kurt, could you please add components next time you assign the bug?

Thank you very much!

Not sure how we want to reproduce this under test, but to reproduce this locally I had a script that generated n number of .html pages, each one with a javascript trigger to load the next one (Rohit's idea!), and then I just browsed to the first page and let it run. IIRC to do about 4,000 navigation events took maybe 15 minutes. 

I'm happy to help here however possible.