The low-level MT rotator controller reports the TRACK enabled substate command as finished before the rotator is actually in the right state to receive TRACK_VEL_CMD commands.
Also please consider increasing the time limit between then the rotator goes to "tracking" enabled substate, and when the first tracking command arrives.
Once a tracking command arrives, it is important to keep sending them. But until the first tracking command arrives, the system is not moving, and so I would argue is not doing anything risky. We can afford to allow the client to wait a few seconds.
I mention this because I tried to work around this bug by adding a 0.5 second delay after sending TRACK enabled substate command, and before sending the first TRACK_VEL_CMD. That caused the system to fault. I ended up using a 0.1 second delay, but it feels to me like the window of allowed sleep times for working around this bug is too small.