Rewrite `ci/style.rs` using `syn`

[tgross35]

The style checking script has been about the same since it was introduced in 2015, but its text-based parsing seems to make it get in the way a lot of the time.

Ideally it would be rewritten using syn, which should be more accurate and not have problems e.g. handling newlines or recognizing when we are within cfg_if. It could become part of the libc-test crate, so cargo run libc-test checks style.

[tgross35]

One of the problems that should be fixed is that currently style.rs forbids >1 s! macro in the file. Usually this makes sense but we should be able to allow >1 if annotated with a #[cfg(not...)] (specifically not) that is unique to the file. Examples:

Accept this:

// Main `s!`
s! { /* ... */ }

// These are not supported on a single arch. It doesn't make sense to
// duplicate `foo` into every single file except one, so allow this here.
#[cfg(not(target_arch = "foo"))]
s! { struct foo { /* ... */ } }

// Similar to the above, no problems here
#[cfg(not(target_os = "illumos"))]
s! { struct bar { /* ... */ } }

Reject these:

// Main `s!`
s! { /* ... */ }

// A positive (no `not`) config: reject because this should go into
// the relevant file.
#[cfg(target_arch = "foo")]
s! { struct foo { /* ... */ } }

// Main `s!`
s! { /* ... */ }

// Similar to the first (accept) example
#[cfg(not(target_arch = "foo"))]
s! { struct foo { /* ... */ } }

// Reject this: same `cfg` as above, should be combined
#[cfg(not(target_arch = "foo"))]
s! { struct bar { /* ... */ } }

If anyone has any interest in picking this up, I'm happy to help.

[tgross35]

Hm, maybe it doesn't need to be limited to only not config, it seems like it would still be nicer to have #[cfg(any(target_arch = "foo", target_arch = "bar", target_arch = "baz"))] than to duplicate the same struct in three different files. Guess that can take some thought once a rewrite is done.