200 reviews/hr is pretty much the limit of usefulness for me, at least. If I'm going to do news in a day, I'll usually mix them in with the reviews so I'm not sure how to measure time spent on those.
My philosophy is that every time you see a card and easily get it right, that was essentially a waste of mental energy: you didn't deepen or strengthen your understanding of the concept, and it would have been better for you to see the card tomorrow, or a week later, etc. If you apply that principle to 1000 cards/day, cumulatively that's a lot of wasted time. I suspect that if you are blasting through 300 cards/hr and really honestly do know the answers to those cards, you would be better served by increasing your intervals, and using the time you save to do qbanks.
^This is kind of the key point of "spaced repetition" - to filter out the easy stuff, and allow you to spend more time working on the difficult stuff. It might feel good to crank out 300 reviews/hr with 95% accuracy but you aren't learning, you're just patting yourself on the back. If things are easy, you probably aren't pushing yourself hard enough.
I think there are diminishing returns on Anki cards. I'm sure every Ankier has had the experience where you get a test question where you could easily answer the relevant Anki card but can't apply that information to this new context. Drilling 2000 cards/day likely just makes this worse. Just my opinion though.