I actually worked on this project. This is just an issue with how browsers give back the parsed data (which is what it is reporting, rather than the authored styles). This is covered in the FAQ - https://developer.microsoft.com/en-us/microsoft-edge/platfor...
I work on a major browser. The key thing to keep in mind is that modern javascript engines have an optimized JIT. That means that they are able to optimize code paths for the most common paths on the web. Since its much less likely to see code in the wold that uses the "new Array" pattern, it will most likely be slower than the much (MUCH) more commonly used literal version.
also, microbenchmarks like this are silly, since you probably will never need several million new arrays to be allocated.
having worked within the enterprise, and on open source with Eran (the author), both can be very tiring. When it comes to the enterprise, however, it was easier to fix these sort of things, or at the very least prevent them from happening again in the future. People can be repremanded, teams reorged, etc. When it comes to OS, you have the possibility of human garbage coming in and dumping all over the project with little the no recourse outside of banning (which does not stop them from creating new accounts and starting all over again).
The concept of a troll inside of the enterprise is more or less unheard of compared to what a popular piece of OSS can get