More batches per frame, direct access to the depthbuffer, extended HLSL and lots of other cool things are in my opinion worth a month of work. It however should not replace DirectX 9 but should be usable alternatively, which could also lead to a clean abstraction needed for OpenGL ES 2.0, which then also shouldn´t be too much work, except the shader system, but then one could maybe just assign two effects to a material and the correct one is then used depending on the renderer.
Btw, to the shader haters around here: Did you know that you can´t render anything in OpenGL ES 2.0 without shaders?
The really big problem there however is probably Lite-C... It would have to compile to another or even several other architectures. Or it needs to be parsed and executed at runtime, which however will most probably take quite some speed out of it. So maybe no Lite-C on Android?

I just see more potential in DirectX11 than in Android support. A new WED however is anyways the most important at the moment and some good way to debug Lite-c...
Also, wasn´t there somewhen the request to implement PerfHUD support? Should be easily implemented (I am however not sure if it is a good idea to have it always usable.) und would help a lot if it comes to more complex rendering techniques.