I think this is where you're making wrong assumptions. ReactOS is free and open-source, and from this comes choice. Even if the main project chooses to work on reimplementing marginal or even unwanted things, a fork is always an option. (A fork from an immature project like it still is doesn't make sense for now.) After all, it's known that people don't appreciate bloat. ReactOS wants parity with Windows 2003 and that is a manageable target. Bits from later Windows versions may be added for important applications but it hasn't have to be nor is necessarily wanted all that Microsoft pushed from Vista hitherto. ReactOS doesn't have to cover all that much, really.
Also, an important nitpeek: ReactOS doesn't reverse-engineer. It employs clean-room engineering.
Yeah you're right it's not reverse engineering in the technical sense. It's a lazy term to use because it sounds like the approximation of the process even though it's technically inaccurate. I should get out of that habit. :)
I don't get why people are pushing back on this point. It isn't a criticism of ReactOS. It's just a pragmatic result of writing a clone of another project that itself is constantly undergoing development.
This is why people more often talk about ReactOS being exciting in terms of supporting older software rather than new. Obviously if it ever got to the stage that it had parity with the latest releases of Windows then that would be amazing but the chances of that happening are miniscule.