As @dra27 notes, the roadmap isn’t exclusive; in this era of OCaml 5, it’s entirely possible for different subgroups to work on things in parallel 
And the roadmap is just that: a direction of travel, following certain principles. Dune (née jbuilder) itself started outside of any roadmap, as a side project that accrued momentum as it solved a problem with build systems for OCaml, and found users. And if the roadmap is wrong and/or incomplete, we’ll just change it.
One thing I really like about Dune package management is that it reduces the lockin to dune. The new client features allow dune to build opam packages, which eliminates any pressure on forcing all OCaml users to use dune as the sole build system! I’m currently getting Docker’s widely deployed vpnkit to build using it at the moment, and there is a heady combination of dune, ocamlbuild, makefiles and shell commands all being wrapped by the new dune frontend now. It also means we can stop maintaining the dune-overlays org as it will be unnecessary, and tools like mirage can be much simpler in how they generate code.
There’s some way to go, but I think this is a very positive direction of travel for OCaml. We’ve got both Windows support now (via opam 2.2) and a unified frontend (via the Dune devex), and both are compatible!