Wait for all assemblers to be ACTIVE before sending geometry

We've recently observed a race condition between the assemblers coming up and connected to the geometry device, and the geometry device sending its geometry. This MR is an attempt at avoiding this race condition by waiting (up to 3s) for the assemblers to be active.

@hammerd

src/calng/CalibrationManager.py

@@ -1069,7 +1069,7 @@ class CalibrationManager(DeviceClientBase, Device):
                 for device_id in self._assembler_device_ids]
 
             await wait_for(gather(*[
-                waitUntil(lambda: assembler.state == State.PASSIVE)
+                waitUntil(lambda: assembler.state == State.ACTIVE)
                 for assembler in assemblers]),
                 timeout=3)