Just a guess but maybe they wanted to build this in a way that it would actually get used.
Every operating system has a system wide way of resolving names to IP addresses, and every application uses it. This new architecture of building custom name resolution into individual applications might be easier for them to build, but it's crap.
Mozilla is doing two things:
1. Bundling DNS with the browser (e.g. ignoring system DNS)
2. Using DNS over HTTPS
Mozilla could still do (1) and then use DNSCrypt + DNSSEC internally. Then, it would actually be used, but they'd be relying on existing technology that actually fits the use-case, rather than DNS-over-HTTPS.
For the record, I don't think you should ignore the system's DNS, either.