The generated code would look like `<code>impl <a href="...">Foo</a></code>`
which the plain text matcher doesn't match. But by using the XPATH notation, the
nodes are flattened and we can correctly assert that `impl Foo` does not occur
in the generated docs.

We need to recurse into stripped modules to strip things like impl methods
but when doing so we must not add any items to the `retained` set.

For legacy reasons (presumably), Windows does not permit files name aux.

Instead of finding aux-build files in `auxiliary`, we now search for an
`aux` directory relative to the test. So if your test is
`compile-fail/foo.rs`, we would look in `compile-fail/aux`.  Similarly,
we ignore the `aux` directory when searching for tets.

Instead of hardcoding knowledge about the strip-private pass into the
rendering process we represent (some) stripped items as `ItemEnum::StrippedItem`.

Rustdoc will, for example, generate redirect pages for public items
contained in private modules which have been re-exported to somewhere
externally reachable - this will now not only work for the `strip-private`
pass, but for other passes as well, such as the `strip-hidden` pass.

Make the compiler emit an error if the crate graph contains two crates with the same crate-name and crate-salt but different SVHs.

Fixes #30535

The only way to get a value for a zero-sized type is `undef`, so
there's really no point in actually having a return type other than
void for such types. Also, while the comment in return_type_is_void
mentioned something about aiding C ABI support, @EddyB correctly
pointed out on IRC that there is no such thing as a zero-sized type in
C. And even with clang, which allows empty structs, those get
translated as void return types as well.

Fixes #28766

Resolve them in this order as well.