Supported devices: general instability especially around large-payload requests

We are seeing some requests fail or stall/be extremely slow around large internal endpoints, such as getting the device database.

We should probably introduce pagination on the backend and/or backed-off request polling on the frontend