I think the main trouble with new exporters which people aren't really thinking of is third party plugins and behaviors. There are already over 100 topics in the Plugins forum, and think about a few years down the line, especially if C2 increases in popularity. Imagine there being 1000 third party plugins. It's great that C2's extensibility makes it more useful to a lot of people, but on the other hand it causes a disadvantage with new exporters.
Take a new C++/EXE exporter: after a lot of work it may be possible to have an official runtime working. However, what do we do about the 1000 third party plugins? That's years more work if we try to do it ourselves. Then, the Javascript SDK is designed to have an extremely low barrier to entry: you can just pop open Notepad and type some javascript you remember from web page development, and you've got a new plugin. For a C++ exporter, plugins would likely also have to be written in C++. The tools are much more complex for this (e.g. Visual Studio is a lot more complicated than Notepad, especially if you get build errors), and the C++ language is *much* harder than javascript, you need to worry about memory and pointers and such and a great deal of things javascript takes care of for you. Not only that but if you're porting a javascript plugin to C++ double care has to be taken to ensure it's *exactly* compatible, to ensure ported projects actually work the same (otherwise there will be subtle and frustrating differences when you port your game). The question is how many javascript developers are up to this, especially if they're just hammering out a plugin in notepad in their spare time? I anticipate the majority of plugins would not be ported. Then, the majority of existing projects cannot be ported anyway due to their dependence on third party plugins. You can design a project from scratch using just the right set of plugins but one of the strengths of keeping the HTML5-only approach is all plugins automatically work on all platforms without any worry whatsoever.