don't require tokio rt-multi-thread feature

[lunixbochs]

No description provided.

[DataTriny]

Hello @lunixbochs
Please run cargo fmt on your branch.

[lunixbochs]

done

[mwcampbell]

How exactly does this work? Why is it necessary to spawn a new thread, create a tokio runtime on that thread, send the runtime back, and then exit the thread that created it?

[lunixbochs]

Wow, I think I was completely on autopilot. I'd only done basic compile+println testing because existing versions of Accerciser still don't work with Accesskit on Linux due to #241

Sending a handle back and parking the thread on a sleep works much better. I tested this approach further by manually connecting to the bus and verifying some atspi methods work.

[mwcampbell]

This approach looks good to me. I just belatedly approved the CI workflow.

[DataTriny]

The winit example seem to still work properly with the tokio feature on.

[lunixbochs]

I just realized I can move the runtime into the thread on creation, which simplifies the code quite a bit.