SDK seems to have a race condition on watch registration

[pylessard]

When we start a continuous graph with many watcahbles to log, there is often a error reported by the SDK where the server sent a value update before the client was ready for it. Most likely some kind of race condition, to be investigated

Example of log when using the GUI and running a graph

$ python -m scrutiny gui --auto-connect
[INFO] Connected to a Scrutiny server at localhost:8765
[INFO] A new device is connected and ready. Session ID: 08b2e45d7d85418a848435faf5ef17ad
[INFO] Server has loaded a Firmware Decription with firmware ID: 71c3631628caffd4f01791828a3cb016
[ERROR] Got watchable update for unknown watchable 306
[INFO] Device is gone. Last session ID: 08b2e45d7d85418a848435faf5ef17ad
[INFO] Server has unloaded a Firmware Decription with firmware ID: 71c3631628caffd4f01791828a3cb016
[INFO] Disconnected from server at localhost:8765