I only test libloot for x86 Windows, x86-64 Windows and x86-64 Linux in CI.
Ideally there would be a way to tell cargo-vet to ignore dependencies
that aren't relevant to a given set of targets, as having them exempt
like this could hide the need to audit them if they do become relevant
to those targets in the future, but without the supression `cargo vet
suggest`'s output is just too noisy.
Using the latter doesn't really add any value, and while libloadorder still pulls in windows, it means less duplication due to version mismatches in the dependency tree.
They're largely impractical to audit, being a huge number of generated API bindings, plus some prebuilt binaries, and they're the official Windows API bindings from Microsoft, so are hopefully trustworthy.