I think in How to model this constraint in a better way? we already reached the conclusion that it is not CVX taking forever, but the solver, and we will debug it from there.
No need to open three threads on the same topic.
On the other hand it is good to provide as much information as possible and not code snippets removed from context. It is much easier for the readers who are trying to help you to ignore the extra information they don’t need, than for them to try to guess or dig themselves to the information they do need.