I was under the impression that n-of-m ecdsa was outlined some time ago by this paper:

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.67.9913&rep=rep1&type=pdf

Perhaps I have missed something? Your technique does however appear a lot simpler for the 2-of-2 case specifically, which is often useful for various contract protocols.