I want to study the fast attack behavior. In the guide, the following picture is provided.

The picture above describes the AGC state machine.
1)It is claimed that AGC needs a configurable amount of time to be held in reset via the parameter attack delay_us. I assume that thisvalue is expressed in microseconds, not in AGC clocks. Is this assumption correct?
2)In the figure above, there are 2 SLS delays plus 5 AGC clocks (not configurable) between 2 gain update counters. Does this mean that after a gain update I have to wait 37 AGC cycles before the AGC becomes active again, assuming SLS delay=16 clocks?
3)Are the SLS delay and 5 agc clock cycles applied during the first attack or do they take effect after the first attack event?
4)After a gain attack, AGC is held in reset via SLS delay while the detectors are held in reset via the peak_wait_time. Do these two timers start at the same moment, i.e when the attack occurs?