that's a good point. i believe in some ways it's aware, but in some ways it's not, and it's those "not" ways that would be a problem.
it's mount namespace aware in that the emulation executes in the same mount namespace as the program invoking it. so programs run inside the container would use the tools inside the container. that's good.
it's not mount namespace aware in that there's a single registry and handlers created inside the container affect programs running outside the container. so it could be used by code inside the container to break out and into the active VM environment as root. while that's not exactly our security boundary, it's still not something we want to expose, if even just by accident.
i updated
https://crbug.com/1193426 with some info, and based on this, i'd say the schedule is "not for a while". or at least, "not until someone super motivated made Linux upstream support this for everyone".
-mike