White wins
The last one is not much of a study. Besides the first 3 moves it's all tablebase stuff and I am sure the composer had no part in it.
Not all 6/7 piece endgames today come from tablebases. Composers still do the essential parts themselves and use engines only for testing. But as soon as things become "boring" then you know that the author has left to mow the lawn!
There is some chance that this study is unsound due to possible winning 8-man EGTB position, i.e | KRNN vs KRNP
That's a good point but we should still give the author the benefit of the doubt. I don't expect an 8-man database to be created soon as it requires thousands of times more computing power. May have to wait for the quantum computer. Or do you have different information?
Obviously, it may be possible to find 7-men transpositions for specific diagrams like this one and then cook them. If anyone is interested in tackling them one by one, that is.