We have discussed defining a GLU ES, but haven’t felt much demand for it yet.
Porting the desktop GLU code to run on OpenGL ES would be a nontrivial task, since it uses evaluators, fine grained vertex calls (glBegin/End), and dynamic glGets, none of which are supported by OpenGL ES 1.0. Still, it might make an interesting project for someone so inclined. Porting it to work with the Common-Lite profile might be close to impossible, as the tesselation code in particular uses lots of floating point.
Implementing GLU that properly preserves the GL state (as the MESA implementation does) is not only difficult, but impossible with OpenGL ES 1.0 (no dynamic gets, and vertex array data needs to be modified). A version that leaks state (although possibly in a documented fashion) is of course doable.