I have also been using SFML for some time, and would hate to see it die. I am a bit of a lurker here on the forums, but I would definitely be willing to help out with SFML. I love the idea of a committee or some guidance group that would make community contributions much more straightforward.
I think that for SFML to really compete with SDL or something with a backing by valve, it should try to be as feature rich and stable as possible, while being actively developed and tested by the community. SFML has gotten to the point where two or three people can not maintain and enhance it in their free time.
However, we should not try to be another SDL, just in C++, but should grow and develop in the way that the community sees fit. This library has MASSIVE potential with its great design and clean API, but I do agree that it seems to be a bit of a "gateway" library with its current feature set.