Or are you saying that because elite bowlers are invariably better than their teammates, it is therefore invariably preferable for elite bowlers to be good SR/bad ER? I guess that makes sense.
Exactly.
I've mathed out a simplified example here to illustrate.
We are assuming 2 bowler attacks, better bowler is A, averages 20, worse bowler is B, averages 30
Scenario 1: Both strike evenly ( 60 SR for both )
They take 2 wickets every 20 overs, giving up 50 runs in that time.
Total: 250 runs, 100 overs bowled ( 50 each )
Scenario 2: Bowler A is strike ( 30 SR ), Bowler B stock ( 90 SR )
Every 5 overs bowler A will take a wicket, and collectively they'll take 4 wickets every 30 overs, giving up 90 runs in that time.
Total: 225 runs, 75 overs bowled ( 37.5 each )
As you can see, by having complementarity between your "strike" and "stock" bowlers, you saved 25 runs over the course of the innings, not a trivial amount. Plus, you also finished your innings much quicker, reducing bowler fatigue.
This outcome is intuitive, and Test teams already play in a way to maximize complementarity. Think of all the times you will see 2 bowlers bowling from opposite ends with different ( 1 attacking, 1 more defensive ) field sets. Ultimately, all of the bowlers we are talking about when we are discussing ATG comparisons ( or even usually ATVG comparisons ) are of bowler set A, not B. Thus a lower strike rate is always preferred for them.