In CMIP6 we managed the experiments with a set of user-mods in order to have the correct list of output fields and frequencies. Because of the complexity of the experiments the structure was pretty complicated.
For CMIP7 we don't plan to manage it this way. The plan right now is to have a few user-mods for coupled experiments that will do a "superset" of all the fields needed. So this will be a user-mod at the CESM level.
Hence, we should remove the CMIP6 output usermods since they won't be the basis for CMIP7. It's also not clear if these usermods and compsets were that commonly used, so some complexity that will be good to remove for usability.