resolve mingw-related build issues by updating windows crate

[j-n-f]

While working on #151, it was necessary to upgrade the version of the windows crate to resolve mingw-related issues. Unfortunately this causes problems in winrt builds:

error[E0277]: the trait bound `IClosable: From<windows::Devices::Midi::MidiInPort>` is not satisfied
   --> src\backend\winrt\mod.rs:164:47
    |
164 |         let closable: IClosable = self.port.0.try_into().unwrap();
    |                                               ^^^^^^^^ the trait `From<windows::Devices::Midi::MidiInPort>` is not implemented for `IClosable`, which is required by `windows::Devices::Midi::MidiInPort: TryInto<_>`
    |
    = note: required for `windows::Devices::Midi::MidiInPort` to implement `Into<IClosable>`
    = note: required for `IClosable` to implement `TryFrom<windows::Devices::Midi::MidiInPort>`
    = note: required for `windows::Devices::Midi::MidiInPort` to implement `TryInto<IClosable>`

error[E0277]: the trait bound `IClosable: From<windows::Devices::Midi::IMidiOutPort>` is not satisfied
   --> src\backend\winrt\mod.rs:266:45
    |
266 |         let closable: IClosable = self.port.try_into().unwrap();
    |                                             ^^^^^^^^ the trait `From<windows::Devices::Midi::IMidiOutPort>` is not implemented for `IClosable`, which is required by `windows::Devices::Midi::IMidiOutPort: TryInto<_>`
    |
    = note: required for `windows::Devices::Midi::IMidiOutPort` to implement `Into<IClosable>`
    = note: required for `IClosable` to implement `TryFrom<windows::Devices::Midi::IMidiOutPort>`
    = note: required for `windows::Devices::Midi::IMidiOutPort` to implement `TryInto<IClosable>`

It's not clear why this is happening, but this work seems separate from solving the re-entrant locking issue in #150 .

[j-n-f]

@Boddlnagg do you know why this is happening only for winrt? Just based on the types involved, I'm guessing

1. midir needs an update to Cargo.toml so that the right features are added for the winrt feature, or
2. this is an issue that needs to be resolved in the windows crate

[j-n-f]

All right, this was simpler than I thought. We no longer need to turn the midi ports into IClosable to close. The windows crate has wrapped everything up so we can just call .Close() on the midi ports themselves.

[Boddlnagg]

In what way does this relate to the build failures (timeouts) in #146 ? (The .close() calls have been adjusted there already.)

Edit: Is that exactly the hangs that are fixed by using reentrant mutexes? Edit2: No that cannot be, because the reentrant mutexes only affect the WinMM backend ...

[j-n-f]

Hard to say. The windows crate is so complicated that when I update it and my issue is resolved, I'm not too concerned with why. I was upgrading windows to resolve an issue with the mingw build. Debugging the issue there isn't worth my time.

I think the changes now in #151 would also resolve #146. It seems the build hangs for 0.54 (#146), but not for 0.56 (#151).

#151 is ready to merge, and all tests/builds are passing (I got that fixed today).