Define conventions for hand tracked avatars

For now, we assume that users with avatars only use controllers. Both hands of the avatar description in the shared state describe the translation and rotation of a controller. This is limiting for use cases that can use hand tracking. Especially, this is already an issue for the SubtleGame (see ) where the recording is not aware of the hands even though half of the tasks rely on hand tracking. We can imagine this to be limiting for nanover-imd in the near future.

I raise the issue here as it is a central point for the project, but the work won't happen formally in this repository. The steps for fixing the issue are:

* [ ] decide on a convention
* [ ] describe the convention in the documentation

Once the convention is established:

* [ ] implement avatar hands in nanover-imd
* [ ] implement avatar hands in SubtleGame

A reasonable description could be based on the OpenXR specification. If the hand description in Unity differ from that, it could also be a possible base.