Invoke a DeviceLostClosure immediately if set on an invalid device.

[bradwerth]

Connections
N/A

Description
This handles another case in device_set_device_lost_closure where a DeviceLostClosure is given to wgpu but never called. The contract with these closures is that wgpu must call them before they are dropped.

Testing
This adds a new test DEVICE_INVALID_THEN_SET_LOST_CALLBACK which invalidates the device and then sets a device lost callback on it. Unfortunately, there was no existing way to make a device invalid other than dropping it, so it also adds a function make_invalid to replace the device with an error in the registry. A few functions had to be touched up to handle this unexpected error object in the registry, representing an invalid device.

Checklist

• Run cargo fmt.
• Run cargo clippy. If applicable, add:
  • --target wasm32-unknown-unknown
  • --target wasm32-unknown-emscripten
• Run cargo xtask test to run tests.
• Add change to CHANGELOG.md. See simple instructions inside file.

[teoxoy]

Looks good! I left a few suggestions, I can apply them if you think they are good.

[ErichDonGubler]

Pushed up bb03319 to abbreviate assert!(a == b, …) to assert_eq!(a, b). assert_eq generally provides superior diagnostics, and, even better, it means you don't have to write your own failure text. 😀

[ErichDonGubler]

Pushed 73d90fb to further clean up tests being added with deriving Eq on DeviceLostReason, so we can leverage assert_eq!(…) s'more.