I am unsure what version of Stockfish you were using so I ran analysis 12 times; 4 seperate engines for each version of Stockfish.
- Stockfish 10 vs ( Ethereal , Komodo, LC0 , Stockfish 12 )
- Stockfish 11 vs ( Ethereal , Komodo, LC0 , Stockfish 12 )
- Stockfish 12 vs ( Ethereal , Komodo, LC0 , Stockfish 12 )
Results were Stockfish (white) at the end of 20+ moves, was in a stronger position in all 9 instances. Based off this, I would suggest black is in a losing and in a irrecoverable position. There is the potential for a draw but it appears extremely unlikely.
The best variation / playout that I saw, that had the best end position for black, was Stockfish 12 vs Stockfish 12. +-3.03 / 41 so white is decisively better.
Issue with engine games in these sort of positions are the gameplay is boring. That being said, I have attached it for reference though.
[White "Stockfish 12"]
[Black "Stockfish 12"]
[FEN "4b3/4N1k1/3R3p/r7/8/P6P/5KP1/8 b - - 0 1"]
1... h5 2. Rd3 Kh6 3. Kg1 Kg7 4. Rc3 Kf7 5. Nc8 Kf6 6. Nd6 Bg6 7. Kh2 Ra4 8. Rf3+ Kg5 9. Nb5 Kh6 10. Nc7 Ra5 11. Ne6 Bf5 12. Re3 Bg6 13. Rc3 Bf7 14. Nf4 Kg5 15. Ne2 Ra4 16. Rf3 Bg6 17. Rg3+ Kh6 18. Rc3 Be4 19. Nf4 Ba8 20. Ne6 Ra5 21. Nc5 Bd5 22. Nd3 Ra4 23. Nb4 Be4 24. Re3
From my experience with engine games this is going to continue on for some time. Eventually the Black h pawn and Whites a pawns will go (will likely also involve trading out the bishops somewhere ) , white with the pawn advantage will promote.