I'm running this locally to see if the failures for CUDA are related or not. We'll also need a TO tag first.

Can't repro the factorization failures, annoying...

Now needs the MAK bump which itself needs the Mooncake and Enzyme ones

Codecov Report

✅ All modified and coverable lines are covered by tests.

... and 13 files with indirect coverage changes

OK I think this is good to go now that things are passing

... or not. These chainrules failures are suprising to me, it seems it should be unrelated as this PR doesn't touch that code. Could it maybe be related to the MatrixAlgebraKit version bump, as a lot of the failures are in factorization related tests?

Actually, it turns out that the chainrules is the same issue that should be resolved here: QuantumKitHub/MatrixAlgebraKit.jl#219 (at least, I assume so since it only affects the LQ decomposition)

Given that this failure is unrelated, if the tests pass we should probably try and merge this.

Looking at the logs, all the failures are in the LQ tests mentioned above, so I am going to force-merge