> Starship has absolutely undergone extensive testing and continues to do so
That is the Starship module part. I edited out a portion where I indeed incorrectly claimed that Starhsip module hasn't been tested, and replaced by pointing out that the combination of launch vehicle (super heavy booster) + starship hasn't been tested. When comparing "Starship" with SLS, we actually mean the whole two-stage rocket, not simply the part of it named Starship.
> The next Artemis doesn’t launch for something like 2 years. Starship/Superheavy will be flying by then.
This remains to be seen. It was supposed to be flying this year, or even last year in some older estimates. As we've seen from SLS itself, (6) years of delay for these types of subjects are not surprising.
> SpaceX’s refueling for approach may need iteration but they have shown incredible ability to iterate.
Absolutely, but iteration necessarily implies changes in the current design - which is why I'm saying we don't actually know what the final version of Starship will support.
> If you were running a business and deciding on a 5 year vendor contract, would you just ignore the vendor who had a 10x better product but required you to wait a bit (and you were not in a rush?)
It depends on how confident I can be, and in how little of a rush I am. If I think they will deliver the first copy of the 10x better product 5 years from now, I may well award a 5 year contract to the vendor of the 0.1x product today, and have something to work with for the next 5 years.