The OS X kernel has RT facilities. But you're missing the point: you don't need RT for this. The reason why RT is used in Linux for audio just shows the problems the audio stack has. RT is not needed, unless you're doing something wrong in the audio infrastructure.
From experience I know that a task like this can not be taken lightly. I've been developing a lightweight 3D audio library for years and only recently it got in a state I find acceptable (but maybe I've got too high standards). It's called AeonWave and is available for Linux and Windows now: http://www.adalin.com
Although it is not Freeware, it is cheap and a royalty free (but limited) version of the library is available.
I am a fan of Free and Open Source software though (hence the many open source side projects and more to come).
This project was started out of frustration for the state of OpenAL-Sample years ago (and in some ways still today with OpenAL-Soft) and renders about 6 times faster than both. As time passed proper stereo mixing and effects supports has been added as well as support for audio-frames which can be used for sub-mixing and to move multiple sound emitters around simultaneously in a 3d environment.
This library however is not meant to be a replacement for ALSA, PulseAudio or OSS but instead it simply uses one of them as a backend. I really wouldn't want to think of this much code added the the kernel..
Personally I think the library fills a gap that both OS-X and Windows already have a solution for and the fact that it can render (3d) audio up to 7 times faster (and I see some room for more improvement) and can be setup for latencies as low as 2 msec means it is well thought out in my honest opinion.
Sounds good architecture decisions except for the use of OSS.
A company tried to blackmail Linux with OSS API, that's why it is being replaced by ALSA.
The company simply did not release some improvements. People were free to implement those on their own. The creation of ALSA was far more than was necessary and wasted resources that could have went into making other things better.
NASA launches a rocket carrying a satellite. The rocket explodes 10 seconds after liftoff. 1 billion dollars wasted. Someone says: "It exploded, something went wrong!" 89c51 says: "Explain what's wrong then." Someone answers: "I've no idea!" 89c51 says: "See? Nothing's wrong. Everything is OK."
89c51 was fired soon after.
The differrence is that the audio stack doesn't explode or destroy 1 billion dollars of stuff. You dodged the question because you actually know what the answer is: there is nothing wrong with the linux audio stack. Having Real Time priority will allow any type of process to get more done in a given time frame, so it naturally helps audio stuff which has a strict deadline. That being said, it is prefectly possible to do audio stuff on a stock kernel. I generally have between 8 and 32ms latency, depending on how much processor intensive stuff I am doing. I use the stock kernel as I can't be bothered to install a RT one. Judging from the conversations that I have seen on #opensourcemusicians on freenode (people who have some idea how audio works under linux :P) using a RT kernel isn't even the most common approach.
Next time you say something is broken at least be able to tell why you believe this (the rocket because it exploded, a car because it crashed, etc.)