Add endpoints for getting and setting workspace user roles

[jjnesbitt]

Prerequisite functionality for multinet-app/multinet-client#107

This would be a pair of endpoints that would look something like:

• GET /api/workspaces/{workspace}/permissions
• PUT /api/workspaces/{workspace}/permissions

The data model for each of these would contain a list of users, and their permission on the workspace. Something like:

[
  {
    user: <sub>
    role: <'reader' | 'writer' | 'maintainer' | 'owner'>
  },
  ...
]

The workflow for modifying the permissions of a workspace would be as follows:

1. Retrieve list of users with their respective roles
2. Modify list
3. Call PUT to set the new permissions

This would obviously be gated behind the owner role on a workspace.

[waxlamp]

Instead of a list of individual users, would it be better or worse to have something like:

{
  owner: <sub>,
  maintainers: [<sub1>, <sub2>, ...],
  writers: [...],
  readers: [...]
}

and you would modify the permissions structure just by adding/deleting users from the appropriate lists.

What are the pros/cons of doing it each way?

[jjnesbitt]

Instead of a list of individual users, would it be better or worse to have something like:

{
  owner: <sub>,
  maintainers: [<sub1>, <sub2>, ...],
  writers: [...],
  readers: [...]
}

and you would modify the permissions structure just by adding/deleting users from the appropriate lists.

What are the pros/cons of doing it each way?

The approached you laid out is preferred, as we already model the permissions in this way anyway. I'll follow this approach.