investigate poor benchmark results

[iapicca]

round 21 of techempower's benchmarks' results are out json serialisation single query multiple queries fortunes data update plain text angel3 seems to be much worse than plain dart, maybe there is some bottleneck worth investigating

[dukefirehawk]

@iapicca Thanks for the update on the performance benchmark. Version 4 is basically the first release after we migrated the entire codebase to null safety whereby the primary focus was to get everything working correctly first. Now that is pretty much settled, we can now look into ways to crank up the performance.

[iapicca]

good thing that next round is quite early, next october or november