I wish they would consider using cacheline sized B+ trees instead of dumb RB trees.
The latter are not making proper use of pipelined superscalar processors (AKA any modern CPU that can run at over 1 GHz).
Additionally, one must learn the politics, code-style, idiosyncrasies, etc before submission will be successful. And of course, the architecture of the project itself.
Open Source / FOSS creates the opportunity for anyone to offer code; it does not mean it will so simply be accepted, or should be. And it does not mean you can, or should. But if you wish to, a path always exist (if they just plain don't want it, and you really want to add it... fork!)
Coding data structures in C is a formula for wasting your time, because at each next use you have to start over nearly from scratch. That is why kernels are such heavy users of ancient data structures user-space has largely abandoned.