Right now we are pretty much as equaly frustrated with Ludei and CocoonJS as you guys are. It's not any easier even for us to get through to them, and there are some basic problems like memory management that have gone unaddressed for months. We're actively looking in to alternatives. However I must say it is actually nearly there. It's not terrible, and if they fixed memory management, the issues with audio, and a handful of other noticable bugs, then they'd have a solid platform. The problem is they seem to be struggling to achieve this and we're all getting impatient waiting.
We are strongly opposed to developing our own engine (i.e. not a real browser engine). We think the basic approach is wrong: there are so many browser features we use that it becomes a fool's errand to try to implement everything. How many of these non-browser HTML5 engines do you know that support the Web Audio API, WebRTC, form controls, XML parsing and more? We'd rather invest in some solution using a real browser engine. And hopefully something based on this approach will turn up soon, since we know a couple of big players are working on projects in this area. If nothing turns up soon, we'll look in to it ourselves.