That's exactly what I did with the LE.NET, but only a few used it.
I suggested that (exactly as you do) the Leadwerks .NET guys should use it as a base for their more specialized and extended ideas. To make a long story short. They didn't. By now I'm feed up with working on language wrappers that get flushed out into space, so I have left that behind and will use C++ and Lua. However ZioRed still supports the LE.NET I think and he's a great guy and understands the idea with LE.NET.
I totally agree with Josh, 100%.
First make a 1-1 wrapper which acts as a stable base. Then use that for all other experiments and ideas. I said it back then and still think this is the correct way to go.
Hopefully someone that can accept this model of achieving stable results will continue supporting the LE.NET idea. ZioRed has done this since I ended my work with that. Maybe he would be the guy, that he has to decide him self.
Anyway I wish all C# efforts good luck.
Roland