Should the security and performance issues be also addressed for the WebRTC+JavaScript implementation to become a full featured one, I think some nodes might also serve as bridges between this implementation and the native one. Although the current C++ implementation would need be extended for that, I see no technical reason why the original design would not support it. People could then try the Safe network without having to install anything and later install the native node, should they want to.
I don’t think it should be put on the critical path for the launch of the network and we should wait for the results of my initial experiments before prioritizing in any way. However should that work out, the adoption speed for the Safe network would only be limited by social factors and not technical ones, since the design does not require resources others than those provided by its users.