The core developers absolutely do believe the market should decide upon the blocksize limit and direction of the codebase. I have never heard the argument from one of them that their should only be one implementation and people should be restricted to only using their repisitory.
I recall some discussions a while ago in which one core developer defended a single reference implementation as defining Bitcoin. As I recall, there were two separate discussions, one why the alternative approach used by the IETF (protocol specifications) was not appropriate to bitcoin and another indicating the risks of network forks due to alternate implementations, e.g. btcd. I have not seen any change in this viewpoint in the form of public documents or forum posts.