Between O1 and O2, I had added automatic checkers to several Initial alignment states that give the operator a notification when the WFS have converged and are ready to be offloaded.  I think these are working well enough to be relied on, so that operators can just request the Offloaded states and skip the states where you wait and check that the loops have converged.  To encourage everyone to use these checkers, I have taken states that have them out of the list of requestable states in ALING_IFO. 

This means there are now three fewer steps in the Inital alignment procedure that are done by hand: INPUT_ALIGN, SRC_ALIGN, and PRM_ALIGN can all be skipped and the operator can simply request INPUT_ALIGN_OFFLOADED ect. 

I have also done a two things that I think should make MICH_DARK more reliable, I fixed a big which meant that it did not always wait for the suspensions to stop swinging before trying to lock, and I increased the input power for MICH locking to 10W which should make it less sensitive to dark offsets.