So really there are only "more selective" and "less selective" drugs.
Agree that is true - but I was getting at a concept. In your more realistic space I would still agrue that having a complete library of substantially "more selective" drugs is likely to be more efficacious than one cover-all-bases drug. But of course it depends on how much more selective and how comprehensive the library.
Now maybe you can get lucky and find two (or more than two) selective drugs that don't have overlapping off-target effects and end up with a magic combo that is selective for just the two targets you want. I agree such a magic combo would indeed be better than any individual drug as you can optimize the doses independently. But at some point you are also going to break the bank in terms of costs.
We've apparently found 1 - yes? Why wouldn't finding a second and a third happen just as easily? In fact I would suggest that finding the 2nd and the third could actually be easier - because of the incentive to plug the hole that the cancer escapes out of when treated with the first drug.