In order to run performance test, or to provision test environment, we need to be able to start an instance with a bench of pre-configured data.
For Marketplace Packages and Connect Client, we don't really need real data user.
Complexity is not tied to a specific client, but of the multitude of possible edge cases between clients.
We prefer to generate fake data with managed cases rather than export existing production data.
For (git) project, it's hard to mock a complex project; reproduce real branches complexity, amount of data, ... Thus, we need to rely on a real project without personal data for security reason.
And in order to properly use dev environment, we need to hydrate the server with some relevant packages: nuxeo-web-ui, nuxeo-jsf-ui, ...