The main intention behind using the openshift-connector branch was to develop a new feature within the main codebase (so that all committers can review/comment) while not interfering with the release schedule of mainstream Che. The other options for Openshift integration, as I see it, are 1) to make PRs on master for all changes, or 2) to develop Openshift integration in a fork of Che.
For the first option, Openshift integration is not yet at the point where full release makes sense, so we would either be introducing a large support burden with each release, or adding code without release notes that risks causing instability for users who don't want to touch Openshift. It's just not a feature that can be developed over the course of one release cycle. We'd also waste time documenting changes that may be removed in a later commit.
For the second option, moving to a proper fork of Che would do nothing to solve the problem of eventual integration while also obscuring the development process and discussions that take place.
FWIW, I don't think a branch on the main repo is ideal, but it seems to be the best option we've got at the moment. Maybe it would help if we labelled Openshift-related PRs with a openshift-specific label, so as to not clutter the list for master? We can discuss more later.