Fix #[func]-like attrs being ignored with macros below them

[PgBiel]

This fixes the bug mentioned in #379 (comment).

To better explain the problem, consider the following modified code from the Dodge the Creeps example.

#[godot_api]
impl Mob {
    #[func]
    #[cfg(all())]
    fn on_visibility_screen_exited(&mut self) {
        self.base.queue_free();
    }
   /* ... */
}

Note: #[cfg(all())] is a #[cfg] check which should always pass / be a tautology (something like #[cfg(TRUE)], but that doesn't work by itself). This PR does not fix (yet) errors which appear when #[cfg] removes the function.

Before this PR: compile-time error:

    Checking dodge-the-creeps v0.1.0 (/.../gdext/examples/dodge-the-creeps/rust)
error: cannot find attribute `func` in this scope
  --> examples/dodge-the-creeps/rust/src/mob.rs:17:7
   |
17 |     #[func]
   |       ^^^^

error: could not compile `dodge-the-creeps` (lib) due to previous error

After this PR: compiles.

The problem, as mentioned in my original response, is that the extract_attributes function would keep re-assigning the found: Option<BoundAttr> variable even after a #[func]-like attr was already found, thus setting it to None when some unrecognized attribute appeared under #[func]. This fixes the problem by just skipping unrecognized attributes - found is only re-assigned if a #[func]-like attribute is actually found (and this preserves the error when two consecutive #[func]-like attributes are specified).

Added tests under itest.

(Also no idea why clippy is failing here lol, I didn't change that file!)

[GodotRust]

API docs are being generated and will be shortly available at: https://godot-rust.github.io/docs/gdext/pr-442

[Bromeon]

Thanks!

Can you reference the symbols somewhere in Rust code (assign constant or function result to dummy value)? This would test that the symbols are indeed present.

[Bromeon]

I'd use "below" instead of "under", and "constant" instead of "const".

Also, can the identifiers be a bit shorter? "should still be recognized" can be shortened to just "recognized".

Same applies to functions 🙂

[PgBiel]

Done, and done! 😁

[Bromeon]

Thanks a lot!

Regarding failed clippy, this is addressed in #441, so I might merge that one first (given readiness).

[Bromeon]

Replaced assert!(ALWAYS_TRUE_CONSTANT) with static_assert! due to Clippy warning. These technically don't need to be inside the #[itest], but it's easier to understand like this.