Allow easy async fn declaration for pyfunction

[ShadowJonathan]

To me, the current examples look very verbose for what they eventually do;

#[pyfunction]
fn sleep(py: Python) -> PyResult<&PyAny> {
    pyo3_asyncio::tokio::future_into_py(py, async move {
        let py_sleep = Python::with_gil(|py| {
            pyo3_asyncio::tokio::into_future(
                py.import("asyncio")?.call_method1("sleep", (1,))?
            )
        })?;

        py_sleep.await?;

        Ok(Python::with_gil(|py| py.None()))
    })
}

To me, this looks easily collapsible into something like this;

#[pyo3_asyncio::tokio::async_pyfunction]
async fn sleep() -> PyResult<&PyAny> {
    let py_sleep = Python::with_gil(|py| {
        pyo3_asyncio::tokio::into_future(
            py.import("asyncio")?.call_method1("sleep", (1,))?
        )
    })?;

    py_sleep.await?;

    Ok(Python::with_gil(|py| py.None()))
}

This would be a little bit more ergonomic to write.

[awestlake87]

Yeah, I was actually thinking about this again recently. I opened PyO3 PR 1406 awhile back to get my thoughts on it out there. The main issue I had with it at the time was that it would have been a bit weird to provide tokio specific or async-std specific stuff in the PyO3 core.

Moving that #[pyfunction] proc macro here would have been a lot of duplicated code and more than I wanted to maintain, but if we can instead delegate to #[pyfunction] from our own runtime-specific proc macro like you've laid out, then I think it would work better.

I think we can do something like this in the proc macro:

quote! {
    #[pyo3::pyfunction]
    fn #fn_signature {
        #entire_async_fn

        pyo3_asyncio::tokio::future_into_py(py, async move {
            #fn_name(#args).await
        })
    }
}

The main question is how much work do we have to do ourselves to get those variables in the proc macro