Refactor opensim-gui to use the opensim-viewer codebase underneath

[aymanhab]

While we can use the javascript code as is to display the static viewer inside the opensim-gui work need to be done to explore:

1. Using gltf in memory to communicate with the viewer
2. 2-way communication with the viewer (as needed by the edit functionality)
3. Handle off-line operation without backend and possibly no python connection

[aymanhab]

There's support to run ReactApp and access it using websockets, the specifics for the openim-viewer and the opensim-gui need to be ironed out. The pros of this approach:

1. Lighter coupling between the gui and the viewer
2. Ease of upgrading and integrating other third party tools that are React based.

[aymanhab]

This issue should be converted to an epic that spans multi-sprints per the dev team plans/presentations:

• Deploy opensim-viewer application and load into GUI without model(s)
• Customize viewer to remove items that don't belong in application (login, gallery, toolbar,...)
• Add options in viewer to parallel the menu in application/GUI
• Add websocket layer around viewer code base and test two way communications
• Convert GUI code to create and pass gltf scene instead of old threejs format
• Remove GUI animation slider and animate in browser
• Wire editing to viewer
• Implement muscles.

[aymanhab]

issue #134