Have climate models been tested, whether if they keep running for a century with repeated input from the historical period would the projection remain stable and ~repeat that regime?

It is regularly done for the GCMs (so called control runs). It is more rare for RCMs but has been done for some of them. Models are usually stable in such tests.