I modified the Guardian to accommodate alignment needs. The OL control engaged state is removed and it becomes a part of the aligning state. In the aligning state, the guardian will engage IM to TM DC control. Instead of referencing a saved reference point, the setpoint will be taken to be the K1:VIS-XX_GOOD_OPLEV_PIT/YAW which Nakano-san created to save OL positions.

Meanwhile, in the misaligning state, the guardian will use the OPTIC ALIGN instead because misaligning requires large offset which means putting the QPDs off screens. Now, the misalign offsets in both PIT and YAW is just a hardcoded 800 counts of actuation which is a value known to work.

I attached the state graph as a reference anyway although it's not so interpretable.