Ah, I thought your claim was that engines can now match correspondence players in the endgame. I believe such a claim would be wrong but I'd love to see it tested!

Yup, engines overall play endgames very strongly, normally overall better than GMs who probably make more mistakes in the endgame than any other phase.

But engines also still make basic errors in the endgame, so even a mere chess mortal like you or me can add significant value to an engine's endgame play.  I can add much more value to an engine's endgame play than I can to an engine's middlegame play.

I feel we are currently at the point with endgames that were at with middlegames a few years ago.   Go back a  few years and engines didn't necesarily play middlegames that well in terms of good postional play / "plans" but were strong anyway because they were already tactical monsters.    Now they are just good in almost every respect at most types of middlegames (some types of KIDs bieng a notable exception as your game with Matt showed).

Engines in my opinion still rely a lot on brute force in endgames, which is fine when they are few panws, but less good with more pawns on the board.   And I think it will be much harder to round them off for endgames than middlegames becasue the nature of the problem is different.