Not sure my thoughts will be coherent, and maybe they can’t be, but love the thread.
Yes, routing and features are two different skills. That said, and from my limited exposure working in one office, heading my own office that did produce a few that went on to success, and having collaborated with a few fellow golf course architects, in most cases, someone artistic enough to be a good router, was also pretty good at features. But, not universally.
Each routing is unique, hardly ever to be repeated again. With features, many have repeated past used features (whether their own or replicas/interpretations of others) and the biggest problem I have seen is really approaching feature design as cut and paste problems, whereas its harder to do with routings, other than on a flat site.
With my own routings, I have tended to judge them most by the number of good holes, further refined by the number that required no earthmoving other than to build tees and greens. Not needing to build many fairway hazards can be a sign of great routing, or at least using everything you have. As time goes on, I even measure the total distance from greens to next tee as one test. A perfect 18 hole course shouldn’t have more than say, 200 feet x 17 connections, or 3400 LF. I will say, I have at least one course with 10X that, but sometimes sites are split and it is necessary, if not good.
I also have listed criteria for good routings, as well. Minimize sun problems, etc. etc. etc. that all have proven to be a big advantage in golf enjoyment. That said, even after applying those criteria to a routing, it is usually the gut feel that prevails in the end, as to which I choose to present to a client. Sometimes, they just “feel right” on plan. For instance, a plan of gentle bends nearly always looks better and more graceful than a back and forth of straight holes, both on paper and in field. But not always.
So, for both aspects, which are about as different as driving and putting in actual golf, I still go back to an inherent instinct towards the creative approach, modified by real world considerations. Basically, you just have to think a certain way overall, plus have received some training for someone who knows what they are doing to be really good at this profession.
Historically, most architects have partnered with those who complement their best talents, i.e. Langford and Moreau, but many others as well. Others simply hire complementary talent, but several kinds are necessary.