kernel arguments and group size thread safety recommendations

The spec doesn't seem to explicitly say this but kernel objects are currently defined should be generated from modules per command recording thread.

I wonder if it would be better to add a way to just record a kernel to a command list, set group size and inputs for and dispatch it. This seems more thread safe and would allow for sharing kernel objects across threads if all you wanted to do was record a command stream with different inputs or group sizes.

Otherwise you'd have to create a new kernel object per command stream, I suppose it depends on whether the API usage envisages kernel objects are being light to construct or something that will trigger a compiler backend pass.

Review the kernel-object and command-recording API described in the issue, focusing on how kernel arguments, group size, and dispatch are expected to work across recording threads. Done would be a clear specification decision and documented API behavior for sharing kernel objects versus creating them per command stream.