Added async I2C implementation

[asasine]

I saw that async I2C wasn't implemented and I'd like to use it for a sensor driver I'm writing. I implemented this following how embedded_hal_mock::spi::Spi was implemented for async.

Closes #118

[asasine]

I don't love how the tests are implemented because there's a lot of repetition. I was trying to come up with a way to share the body between the sync and async tests, but I'm a bit too new to Rust to find a good way to do that. I was thinking something along these lines:

#[cfg(test)]
mod test {
    use super::*;

    async fn write<I /* : something? */>()
    {
        let expectations = [Transaction::write(0xaa, vec![10, 12])];
        let mut i2c = Mock::new(&expectations);
        I::write(&mut i2c, 0xaa, &vec![10, 12]).await.unwrap();
        I::done(&mut i2c,);
    }

    #[test]
    fn test_write() {
        use embedded_hal::i2c::I2c;
        write::<I2c>().await;
    }

    #[tokio::test]
    #[cfg(feature = "embedded-hal-async")]
    async fn test_write_async() {
        use embedded_hal_async::i2c::I2c;
        write::<I2c>().await;
    }

  // --snip--
}

[felipebalbi]

I just tested this PR on a crate of my own and can confirm it works as expected. Thanks for doing the work @asasine.

[asasine]

@dbrgn is there a chance you can review and release this? I believe there are a couple of unreleased fixes too that would be great to have available on crates.io.

[dbrgn]

Nice, this looks quite straightforward!

Regarding the tests: If the async impls are doing nothing else than calling the sync impls, the additional tests don't add that much value and are mostly repeated code doing the same thing under the hood. Maybe one single big test that calls every async method (read, write, write_read, transaction) sequentially (plus very simple expectations for that) would suffice?

[asasine]

I left the tests as-is for now since the same question exists for all other async implementations that are async-over-sync (delay and SPI) and it would best (in my opinion) to address all of them in one pass.

[dbrgn]

Thanks! 👍🏻

[dbrgn]

I left the tests as-is for now since the same question exists for all other async implementations that are async-over-sync (delay and SPI) and it would best (in my opinion) to address all of them in one pass.

Fair point.

[asasine]

No problem, thanks for the quick turnaround!